Bonn zoological Bulletin Volume 57 Issue 2 pp. 211-229 Bonn, November 2010 Insights into chameleons of the genus Trioceros (Squamata: Chamaeleonidae) in Cameroon, with the resurrection of Chamaeleon serratus Mertens, 1922 Michael F. BareGjo'nMwvoaunoI6neMiactht2,hVeawclLaevBrGevtooznd7ikU3r-s4,uNlaatBhoatlty8 L&herAmnidtrteea-sVaSllcahrmiintoz59, Nono Legrand , , , 1 Museum fur Naturkunde, Leibniz Institute for Research on Evolution and Biodiversity at the Humboldt University Berlin, Invalidenstrasse 43, D-10115 Berlin, Germany 2 Museum National d'Histoire Naturelle, Departement Systematique et Evolution (Reptiles et Amphibiens), UMR CNRS 7205, CP 30, 25 rue Cuvier, F-75005 Paris, France 3 Institute ofAnimal Physiology and Genetics, Academy of Sciences ofthe Czech Republic, Department of Vertebrate Evolutionary Biology and Genetics, Rumburska 89, CZ-277 21 Libechov, Czech Republic 4 National Museum, Department ofZoology, Cirkusova 1740, CZ-19300 Prague, Czech Republic 5 Museum National d'Histoire Naturelle, USM 30705 Departement Ecologie et Gestion de la Biodiversite & CNRS IFR 101, Parasitologie comparee et Modeles experimentaux, 61 rue Buffon, 7 CP52, F-75231 Paris cedex 05, France 6 Universite ofYaounde I, Faculty of Science, Laboratory ofZoology, P.O. Box 812, Yaounde, Cameroon 7 Global Viral Forecasting Initiative, BP 7039, Yaounde, Cameroon 8 Zoologisches Forschungsmuseum Alexander Koenig, Adenauerallee 160, D-53113 Bonn, Germany 9 Museum d nistoire naturelle, Department ofHerpetology and Ichthyology, CP. 6434, CH-1211 Geneva 6, Switzerland * Corresponding author: E-mail: [email protected] Abstract. Relationships among chameleons ofthe genus Trioceros in Cameroon are reviewed on a molecularbasis us- ing mitochondrial genes and by morphology. Trioceros oweni is placed basal to two distinct clades (lowland-submon- tane speciesvs. submontane-montane species) anditsposition is discusseddueto high genetic differencestotheremain- ing taxa. Within the lowland-submontane species group, distinct subclades with low genetic differences exist within T. montium and T. cristatus. Differing relationships to previously published results are observed within the submontane- montane species group, resulting in taxonomic changes: Trioceros eisentrauti is grouped with the two T. quadricornis subspecies, showing only low genetic differences, which also correlates with the similaroverall morphology. The taxon is thus assigned to a subspecific rank: T. quadricornis eisentrauti. Within the wiedersheimi-gvoup, the former southern subspecies is elevated to species rank, Triocerosperreti, and two additional species have been distinguished by molec- ular and morphological methods in the former nominate taxon. Trioceros wiedersheimi is restricted to northernmost lo- calities, while remaining populations have been assigned to the revalidated taxon Trioceros serratus (Mertens, 1922). DifferentiatingmorphologicalcharactersforthethreespeciesareprovidedandaneotypeofChamaeleonserratusMertens, 1922 is designated and described to ensure clarification ofits taxonomic status and type locality. Key words. Reptilia, Chamaeleonidae, Trioceros, Triocerosserratus, Africa, Cameroon, phylogeny, taxonomy. INTRODUCTION Only recently, Tilbury & Tolley (2009) provided molec- cn'statas-subgroup inside the Trioceros-group from ular evidence that the two former subgenera (Chamaeleo Cameroon. Laurenti, 1768 sensu stricto and Trioceros Swainson, 1839) of the chamaeleonid genus Chamaeleo as recog- The Republic ofCameroon exhibits a very high diversi- nized by Klaver & Bohme (1986) represent two distinct ty ofchameleon species compared to adjacent countries, and valid genera. For a diagnosis ofthe two genera see especially in montane areas (Bohme & Klaver 1981; Klaver & Bohme (1986, 1992) and Tilbury & Tolley Gonwouo et al. 2006; Herrmann et al. 2005, 2006). At (2009). Klaver & Bohme (1992) additionally provided a present 14 species ofchameleons are known to occur in detailedoverview offormerlypublishedknowledge onthe Cameroon. They belong to the genera Chamaeleo [five Bonn zoological Bulletin 57 (2): 211-229 ©ZFMK 212 Michael F. Barej et al. species: C. africanus Laurenti, 1768; C. dilepis Leach, bosomal RNA gene was amplified using the primers 1819; C. gracilis Hallowell, 1842; C. quilensis Bocage, 12SA-L(lightchain; 5'-AAACTGGGATTAGATACC 1886; C. senegalensis Daudin, 1802], Rhampholeon [one CCA CTAT - 3') and 12SB-H (heavy chain; 5' - GAG species: R. spectrum (Buchholz, 1874)] and Trioceros GGT GAC GGG CGG TGT GT - 3') of Kocher et al. [eight species: T. camerunensis (Muller, 1909); T. crista- (1989). PCR cycling procedures were as described in tus (Stutchbury, 1837); T. eisentrauti (Mertens, 1968); T. Schmitz et al. (2005). PCR products were purified using montium (Buchholz, 1874); T. oweni (Gray, 1831); T. pf- Qiaquick purification kits (Qiagen). Sequences were ob- efferi (Tornier, 1900); T. quadricornis (Tomier, 1899); T. tainedusingan automatic sequencer(ABI 377). Sequences wiedersheimi(Nieden, 1910)]. Trioceros quadricornis and were aligned using ClustalX (Thompson et al. 1997; de- T. wiedersheimi are polytypic, with one more subspecies fault parameters) and manually checked using the origi- [T. q. gracilior (Bohme & Klaver, 1981), T. w. perreti nal chromatograph data in the program BioEdit (Hall (Klaver & Bohme, 1992)], resp. (Bohme & Klaver 1981; 1999). PAUP* 4.0b10 (Swofford 2002) was usedto com- Chirio & LeBreton 2007; Gonwouo et al. 2006; Klaver pute the uncorrected pairwise distances for all sequences & Bohme 1986; Tilbury & Tolley 2009). According to (Tab. 2, Appendix II). We performed neighbour-joining Klaver & Bohme (1997) and Uetz & Hallermann (2010) (NJ), maximum parsimony (MP), maximum likelihood one additional species (Chamaeleo laevigatus Gray, (ML) and Bayesian reconstructions. ForMLandBayesian 1863) is present in Cameroon, but this species has notbeen analysis parameters ofthe model were estimated from the & & listed by other recent authors (Chirio LeBreton 2007; data setusing Modeltest 3.7 (Posada Crandall 1998) and Gonwouo et al. 2006; Tilbury 2010). While some species MrModeltest 2.3 (Nylander 2002), respectively. For the such as Chamaeleo gracilis or Trioceros cristatus show MPanalysis, weusedthe "heuristic search" withthe "ran- a large distribution ranging at least fromNigeriato Gabon dom addition" option ofPAUP* (Swofford 2002) with 10 and the Congo (Necas 2004), five species are regarded as replicates, using the TBR (tree bisection-reconnection) montane endemics occupying restricted high elevation ar- branch swapping option. For the ML tree we used the eas along the Cameroon mountain chain, with T. eisen- PhyML 3.0 computer cluster ofthe Montpellierbioinfor- trauti the most restricted, being endemic to the Rumpi matics platform (http://www.atgc-montpellier.fr/phyml/) Hills in western Cameroon (Chirio & LeBreton 2007; (Guindon & Gascuel 2003). All Bayesian analyses were Gonwouo et al. 2006; Klaver & Bohme 1992). performed with MrBayes, version 3.12 (Huelsenbeck & Ronquist 2001). The exact parameters used for the Pook & Wild (1997) published a preliminary phylogeny Bayesian analyses followed those described in detail by of Trioceros from Cameroon, and we herein provide ad- Reeder (2003). For the Bayesian reconstruction clades ditional and new insights into this species group basedon with posterior probabilities (PP) > 95% were considered additional material. strongly (significantly) supported. Additionally, we used bootstrap analyses with 1000 (forML), 2000 (forMP) and 20000 (for NJ) pseudoreplicates to evaluate the relative MATERIALAND METHODS branch support in the phylogenetic analysis. In all, 49 combined, mitochondrial 16S and 12S rRNA In the morphological analysis measurements follow stan- gene fragments, sequences (Tab. 1,Appendix II; museum dard procedures (e.g.Werner 1902; Mariaux et al. 2008) acronyms see below) comprising 964 bp (lengths refer- and were taken on preserved material with an electronic ring to the aligned sequences including gaps) were ob- dial calliper (± 0.1 mm). All measurements are given in mm tained. One short section (4 bp from the 12S gene) was (Tab. 3,Appendix II). Analysis ofmorphological da- toovariable to be reliably aligned, and was excluded from ta has been performed using PAST software (Version the analyses, resulting in a total of960 bp whichwereused 1 .82b; Hammer et al. 2001). Ifmeasurements (e.g. femur in the analyses. Kinyongia tavetana (AM422414/ length) differed between body sides, mean values were AM422433; Mariaux et al. 2008) was used as outgroup. used. Photos ofliving specimen havebeen usedto analyse Its position outside of Trioceros was demonstrated by colouration patterns. & DNA Tilbury Tolley (2009). was extracted using Qi- Amp tissue extraction kits (Qiagen) and the peqGold Tis- Investigated specimens are deposited in Museum d'his- sue DNAMini Kit (PEQLAB Biotechnologie GmbH) (see toire naturelle, Geneva (MHNG); Museum national d'Histoire naturelle, Paris (MNHN); National Museum, W5'a-gnCeGrCetCaTl.G20T0T9Ta)A.TTCheApArAimAerAsC1A6sTar--L3'()liagnhdt c1h6asibnr;- Museum ofNatural History, Prague (NMP6V); Zoologi- H (heavy chain; 5' - CCG GTC TGA ACT CAG ATC sches Forschungsmuseum Alexander Koenig, Bonn ACG T - 3') ofPalumbi et al. (1991) were used to am- (ZFMK); Museum fur Naturkunde, Leibniz-Institut fur RNA Evolutions- undBiodiversitatsforschung an derHumboldt- plify a portion ofthe mitochondrial 16S ribosomal Universitat zu Berlin (ZMB); Zoologische Staatssamm- gene. Additionally, a section ofthe mitochondrial 12S ri- lung Munchen (ZSM). Bonn zoological Bulletin 57 (2): 211-229 ©ZFMK Chameleons ofthe genus Trioceros from Cameroon 213 RESULTS (1997). The distances of T. eisentrauti to the nominate form T q. quadricornis (0.51%-0.64%) are about equal To date ten taxa belonging to the genus Trioceros have in size to the distances ofthe latterto T. quadricornisgra- been recognized in Cameroon (eight species + two sub- cilior(0.63%-0.64%). The distance ofT. eisentrauti to T. species), but the present phylogenetic analysis is incon- quadricornis gracilior is only moderately higher sistent with this arrangement (Fig. 1). All four used phy- (1.02%—1.08%). These values are clearly within the in- logenetic methodologies strongly agree in the overall traspecific distance range of all included Trioceros topologyand in all cases support the same terminal clades. species. Contrarily, the remaining taxa of this subclade The phylogenetic analyses reveal only a single difference show a much highergenetic differentiation between each (discussed below) for the individual analysis ofthe two other, ranging from 3.18%—5.00%. These other terminal applied gene fragments (not shown), therefore, we only clades correspond to the taxa T.pfefferiand T. wiedershei- discuss the results ofthe combined analysis. mi. The latter hornless taxon is represented by three ge- netically well differentiated clades. Two of them corre- Distances between ingroup and outgroup species averaged spond to the two so far described subspecies, but we find 11.48% (10.58%-12.30%; Tab. 2,Appendix II). Interspe- a further significant split within the populations current- cific distances within the ingroup ranged from ly assigned to the nominate form. 3.21%-6.90% excluding T. oweni. Trioceros oweni is the most basal taxon in respect to all DISCUSSION ingroup taxa, which are grouped within one clade fully supported inNJ, MP and ML, while still strongly, but not Following our molecular and morphological results sev- fully significantly supported in the Bayesian eral changes are necessary among Cameroonian reconstruction (PP: 0.91). The main clade is divided into chameleons ofthe genus Trioceros. The overall number two major subclades with strong statistical support. The ofTrioceros taxa in Cameroon is raisedto eleven and two first subclade includes T. camerunensis, T. cristatus, and already known taxa are revised in their taxonomic rank. T. montium, but their mutual relationships remain Within the Cameroonian Trioceros, T. oweni is the most unresolved. However, partly well supported substructure basal taxon, while the other taxa form two subclades, in canberecognizedwithin the two species, T. montium and which T. camerunensis, T. cristatus, T. montium form a T. cristatus. Trioceros camerunensis stands in a basal lowland to submontane group while remainingtaxaofthe position to T. montium, but this is only significantly second subclade inhabit submontane to montane habitats MP & supported by the reconstruction. Within T. montium, (Pook Wild 1997). Results and required changes will we find a subdivisionintothree only slightly differentiated be discussed below in separate sections referring to the subclades. Overall the uncorrected /^-distances of the relevant species groups. included T. montium vouchers range between 0.00%- 0.75%. Similarly, within T. cristatus a similar subdivision Trioceros oweni (Gray, 1831) (Fig. 2A) into three more distinct subclades is apparent. Here, the genetic distances between the included T. cristatus Trioceros oweni, the type species ofthe genus Trioceros, vouchers ranges between 0.00%-1.28%. is the most basal in respect to all remaining Cameroon- ian taxa (Fig. 1). The value ofuncorrected/^-distance be- Contrary to the firstmajor ingroup subclade, relationships tween T. oweni and the outgroup taxon Kinyongia tave- ofthe species ofthe second major subclade remain unre- tana (12.21%) is within the genetic distance range ofall solved and form a basal polytomy. Nonetheless, all ter- included Trioceros taxa to the outgroup (10.58-12.30%, minal clades in this second major subclade are strongly Tab. 2, Appendix II). However, values ofuncorrectedp- supported and are well distinct regarding the individual distances between T. oweni and remaining Cameroonian branch lengths and bootstrap support for each terminal Triocerostaxa(8.57-10.22%) are significantlyhigherthan clade, mostly corresponding to currently accepted species values in-between the remaining ingroup taxa (see Tab. within Trioceros. The morphologically very distinct tax- 2, Appendix II), and the maximum distance value is on- on T. eisentrauti is grouped together with the two de- ly marginally lowerthanthe minimum distance ofall Tri- scribed T. quadricornis subspecies withuncorrectedp-dis- oceros to the outgroup taxon. Based on molecular data, tancevalues ofbetween0.51%-1.08%betweenthese three Pook &Wild (1997) suggestedthat T. owenimight belong taxa. We found only one haplotype in eachofthe two sub- to a distinct species group, being closerrelatedto T.john- specific taxa, T. q. quadricornis and T. q. gracilior, while stoni, an East African species, than to other western Tri- in T. eisentrauti we uncovered a difference of two nu- oceros. In the past, Werner (1902) grouped T. oweni to- cleotide substitutions in our newly gained sequences in gether with T.johnstoni, T. melleri and T. werneri, while comparisonto thepublished 12S sequence ofPook & Wild T. cristatus, T. montium, T.pfefferiand T. quadricornis be- Bonn zoological Bulletin 57 (2): 211-229 ©ZFMK 214 Michael F. Barej et al. Kinyongia tavetana AM422414/AM422433- owen/ Campo region/Nkoelon E146.15 - camerunensis Mt. Cameroon/Njonji E130.1 montium Rumpi Hills/Mofako Balue E179.18 10 montium Bakossi Mts./Edib Hills E188.19 montium Bakossi Mts./Edib Hills E188.18 montium Mt. Kupe/Nyasoso E180.15 montium Bakossi Mts./Edib Hills E188.20 montium Mt. Kupe E130.5 montium Mt. Kupe E131.3 (Trioceros) */* montium Mt. Cameroon E130.4 montium Mt. Cameroon E131.2 cristatus Rumpi Hills/Mofako Balue E180.2 cristatus Rumpi Hills/Big Massaka E180.7 cristatus Mt. Cameroon/Njonji E131.1 cristatus Mt. Cameroon/Njonji E130.2 cn'status Mt. Cameroon/Njonji E130.3 cristatus Campo region/Nkoelon E150.8 _5i cristatus Mamfe region/Amebishu E146.13 cristatus Mamfe region/Amebishu E150.7 - quadricornis eisentrauti (*)/* quadricornis eisentrauti Rumpi Hills/Mt. Rata E178.10 j quadricornis eisentrauti Rumpi Hills/Mt. Rata E178.11 quadricornis quadricornis Mts. Manengouba E131.5 quadricornis quadricornis Mts. Manengouba E130.9 quadricornis quadricornis Mts. Manengouba E130.10 quadricornis quadricornis Mts. Manengouba E131.8 quadricornis graciliorMt. Oku E130.8 quadricornis graciliorMt. Oku E130.7 quadricornis graciliorOku village E131 .4 pfefferi - perreti Mts. Manengouba E131.6 perreti Mts. Manengouba E130.11 l perreti Mts. Manengouba E130.12 serratus Mt.Mbam E178.3 serratus Mt.Mbam E178.5 serratus Mt.Mbam E178.2 serratus Mt.Mbam E178.4 serratus Big Babanki E188.16 serratus Belo, Mt. Oku [NEOTYPE] E130.17 serratus Big Babanki E189.8 serratus Oku village E131 .16 serratus Oku village E131.17 serratus Lake Oku E130.16 serratus Mt. Oku E130.15 serratus Oku village E131.7 wiedersheimiTchabal Mbabo E91.6 wiedersheimiTchabal Mbabo E178.1 wiedersheimi Tchabal Gangdaba E188.13 0.1 Fig. 1. Phylogram ofthe combined analysis ofthe 16S and 12S rRNAsequence fragments (49 sequences / 960 bp intotal). The star symbol "*" denotes significantly supported nodes. [The values for the internal nodes are as follows (NJ/MP/PP/ML, respec- tively): l:(78/61/0.78/73); 2:(95/98/l.00/96); 3:(99/87/0.99/99); 4:(100/98/1.00/100); 5:(100/93/1.00/100); 6:(86/96/l.00/92); 7:(81/89/0.82/86); 8:(97/67/0.56/94); 9:(100/89/0.97/99); 10:(69/83/0.61/63)]. Bonn zoological Bulletin 57 (2): 211-229 ©ZFMK Chameleons ofthe genus Trioceros from Cameroon 215 longed to a different morphological group. A simple 1961; Wagner et al. 2009b) but further studies on this as- BLASTsearch in GenBank, performing a similarity check pectarerequired. Trioceros montium inhabits the submon- ofsequences, ofthe applied T. owem-sequence identified tane zone ofMt. Cameroon, Rumpi Hills, Manengouba T. melleri (16S) or T. sternfeldi (12S) to show the high- Mts. area and parts ofthe south-western edge ofthe Ba- est similarity values; both again EastAfrican species.Ac- menda Highlands (Gonwouo et al. 2006). At first glance cording to Townsend & Larson (2002) andTilbury & Tol- no morphological characters indicate a separation ofpop- ley (2009), T. melleri is related to T.johnstoni. While the ulations. Hence, we refrain to draw any premature con- only western Trioceros (T.feae from Bioko Island, Equa- clusions at this point. & torial Guinea) in the study ofTilbury Tolley (2009) is placed basal to all other Trioceros. Similarly, Townsend Lowland-submontane clade & Larson (2002) found that all western Trioceros (includ- ing T.feae) studied by them stand as a sister clade to the Trioceros cristatus (Stutchbury, 1837) (Fig. 2D) other members ofthe genus. A similar situation appears in T. cristatus anddistinct sub- Hence, concerning T. oweni our results support the view clades are detectable within this taxon and as in the pre- & ofPook Wild (1997) that TriocerostaxainwesternCen- ceding case, uncorrected/^-distances show only compar- tral Africa are more closely related to each other than to atively low differences between the clades (Fig. 1; Tab. T. oweni. The exactposition ofT oweniremains to be as- 2, Appendix II). Stutchbury (1837) described T. cristatus sessed in future studies with a wider sampling from the from Gabon and since then no further subspecies have whole distribution area ofthis genus. been describedortaxa synonymisedwith T. cristatus. Tri- oceros cristatus is widespread in the lowland to submon- Lowland-submontane clade tane zone from Nigeria to the Central African Republic, Gabon and the Republic ofthe Congo (Klaver & Bohme Trioceros camerunensis (Muller, 1909) (Fig. 2B) 1992; Pauwels & Vande weghe 2008). Furthermore, the species has been reported from Ghana and Togo (see ref- In the past Mertens (1964) classified T. camerunensis as erences in Klaver & Bohme 1992) but, these localities a subspecies ofT. montium based on morphological sim- mustbe regardedwith caution, as theyhave notbeen con- ilarities andzoogeographical affinity, butKlaver& Bohme firmed recently. In contrast to T. montium the species is (1992) reclassified the taxon as a valid species. Our mo- more widespread. Amore detailed analysis ofthe overall lecular results do support close relationships between T. distribution must be applied before any conclusions can camerunensis and T. montium but also confirmed its full be drawn. species status. According to Pook & Wild (1997), T. camerunensis is more closely related to T.feae (not in- Submontane-montane clade cluded in our study) than to T. montium. Trioceros quadricornis (Tornier, 1899)-group Lowland-submontane clade (Figs 2E-G), including Trioceros quadricornis eisen- NEW RANK trauti (Mertens, 1968) Trioceros montium (Buchholz, 1874) (Fig. 2C) Molecular results require changes in the former quadri- Withinthe well supported monophyletic T. montium-clade, cornis-group. Morphological distinctness (body size, distinct subclades appear (Fig. 1; Tab. 2, Appendix II). shape of dorsal crest, number and size ofrostral horns, Buchholz (1874) described T. montium from Bonjongo, lung morphology) between populations from southern Mt. Cameroon. Later Mertens (1938) described a sub- parts ofthe Cameroon mountain chain (Mt. Kupe, Manen- species T. montium graft from Mongonge, on the oppo- gouba Mts.) and northern parts (Bamenda Highlands to site side ofMt. Cameroon. Klaver & Bohme (1992) re- Obudu Plateau in eastern Nigeria) have already been rec- garded it only as an aberrant form and moved it in syn- ognized by Bohme & Klaver (1981). Uncorrected p-dis- onymy with the nominate form. Based on dorsal crest tance values between the taxa quadricornis and gracilior & shape, Perret Mertens (1957) indicated a possible sub- (Tab. 2,Appendix II), indicate averyrecent split andthese species from the Manengouba Mts. but, as in T. m. graft, taxa correspond to subspecies. T. q. gracilior is known Klaver & Bohme (1992) proved the occurrence of this fromthe Bamboutos Mts, Mbulu Hills, Mt. Lefo, Mt. Oku characterto be more widespread. However, Pook & Wild and the Obudu Plateau, while T. q. quadricornis is pres- (1997) mentioned differences in the courtship liveryofT. ent on Manengouba Mts. and Mt. Kupe (Bohme 1975; montium between populations. Differences in colouration Bohme & Klaver 1981; Gonwouo etal. 2006; Joger 1982; are ofimportance in species recognition and may play a Klaver & Bohme 1986, 1992). role in character displacement (Pook & Wild 1997; Rand Bonn zoological Bulletin 57 (2): 211-229 ©ZFMK 216 Michael F. Barej et al. Fig. 2. Cameroonian chameleons (in life): A= Trioceros oweni male (Campo region; photo by J.A.M. Wurstner). B = T. camer- unensis (Njonji, Mt. Cameroon). C = T. montium male; specimen with an aberrant horn shown (Big Massaka, Rumpi Hills). D = T. cristatus male (Nkoelon, Campo region). E = T. q. quadricornis male (Mt. Kupe). F = T. q. graciliormale (Mt. Lefo; photo by W. Bohme). G = T. q. eisentrauti female (Mt. Rata, Rumpi Hills). H = T.pfefferi male (Kodmin, Bakossi Mts.). Bonn zoological Bulletin 57 (2): 211-229 ©ZFMK Chameleons ofthe genus Trioceros from Cameroon 217 We group the morphologically highly distinct taxon T. while the taxon eisentrauti from the Rumpi Hills is giv- eisentrautias adistinct subspecies ofT. quadricornis (Fig. en a new systematic status Trioceros quadricornis eisen- 1), a position already indicated by Pook & Wild (1997). trauti (Mertens, 1968) NEW RANK. In contrast to the But, while in the latter publication and in our 12S-only afore discussed species (T. montium and T. cristatus), any analysis (not shown) eisentraati isplaced as thebasal sis- contact zone between these two allopatric taxa can be ex- ter taxon to the two T. quadricornis subspecies, accord- cluded due to their highly restricted altitudinal distribu- ingto ourcombinedresults (16S, 12S) this is notthe case. tion.As above, low genetic differences suggest avery re- Despite its morphological uniqueness (gularcrest formed cent split presumably connected to the altitudinal range offlaps in eisentrauti and a gular crest formed ofconical shifts ofthe lower-temperature forests up to the mountains scales in other Trioceros taxa), molecular results reveal after the end ofthe Pleistocene Ice Ages (when montane close relationships between these three taxa with values forests in the tropics expandedto the lowerelevations; He- ofuncorrected/^-distances within intraspecific ranges. Val- witt 2004). All three subspecific taxa of T. quadricornis ues ofuncorrected/^-distances between eisentrauti and T. could now represent species in statu nascendi. q. quadricornis are comparable to values between the T. quadricornis subspecies and values are only marginally Submontane-montane clade higher between eisentrauti and the subspecies T. q. gra- cilior(Tab. 2,Appendix II). However, the taxa show a dis- Trioceros wiedersheimi (Nieden, 1910)-group & junct distribution with T. q. quadricornis occurring in the (Figs 3A-G), including Trioceros perreti (Klaver NEW RANK Manengouba area (see above) and T. eisentrautibeing en- Bohme, 1992) demic tothe Rumpi Hills inwesternCameroon (Gonwouo et al. 2006; Klaver& Bohme 1997). All taxa inhabit pris- Further changes are necessary within the former wieder- tine montane habitats, T. q. quadricornis occurring at al- sheimi-group. Klaver& Bohme (1992) describedthe sub- m titudes between 1.800-2.250 a.s.l., T. q. gracilioratal- species T. w.perretifromManengouba Mts. Molecularre- m titudes between 1.800-2.400 a.s.l. and T. eisentrautiin sults however reveal full species status for this taxon, as m altitudes above 1.150 a.s.l., respectively (Gonwouo et the uncorrectedp-distances between T. wiedersheimipop- al. 2006; Pook & Wild 1997). We have located T. eisen- ulations from Manengouba Mts. and populations further trauti on Mt. Rata in the Rumpi Hills only above 1.600 north (Bamenda area, Tchabal Mbabo) are clearly within m a.s.l., hence, it is probably even more restricted in its the interspecific range ofother western Trioceros species altitudinal and overall distribution range than previously (Tab. 2, Appendix II). We thus herein elevate the taxon & indicated. In the original description of T. eisentrauti to the full species rank: Trioceros perreti (Klaver Mertens (1968) had already indicated relatedness to T. Bohme, 1992) NEW RANK. The present distribution of quadricornistaxa accordingto body size and shape ofthe T. perreti covers the Manengouba Mts. and the Bakossi dorsal andtail crests. Bohme & Klaver(1981) emphasized Mts. (Euskirchen et al. 2000; Gonwouo et al. 2006). the similarities of T. q. quadricornis and eisentrauti in comparison to T. q. graciliorandremarked that rostral tu- Regarding the former nominate T. w. wiedersheimi, mo- bercles in eisentrauti might represent reduced rostral lecular and morphological results lead to recognition of horns, which are present in T. q. quadricornis (up to two two distinct clades with uncorrectedp-values within in- pairs ofrostral horns) and T. q. gracilior(up to three pairs terspecific range ofthis genus (Tab. 2, Appendix II). For- ofrostral horns). However, reduction of rostral horns is merly, T. w. wiedersheimi has been thought to occur in also known in T. q. quadricornis and T. q. gracilior Cameroon north ofthe Manengouba Mts. (inhabited by (Bohme & Klaver 1981; Mertens 1968) and Bohme & T.perreti(Klaver& Bohme, 1992), see above). Ithas been Klaver (1981) assumed that reduced horns represent a found along the Cameroon mountain chain (Bamboutos more derived character state. From the genetic point of Mts., Mbulu Hills, Mt. Lefo, Mt. Mbam, Mt. Oku and Mt. view, we are aware that the low genetic differentiation in Tchabal Mbabo) and in easternNigeria (Gotel Mts., Mam- mitochondrial DNAmightbe in some cases causedby in- billa Plateau and Obudu Plateau), where it inhabits mon- m trogressive hybridization inthe evolutionaryhistoryoftwo tane savannas and grasslands between 1400 and 2450 species. However, we believe thatthe similaroverall mor- a.s.l. (Bohme &Nikolaus 1989; Chirio & LeBreton 2007; phology (body shape and size, shape ofthe crests) of T. Dunger 1967; Gonwouo etal. 2006; Herrmannet al. 2006; quadricornis andeisentrauti also furthersupports ourhy- Klaver & Bohme 1992). Nieden's (1910) description of pothesis oftwo closely related, but conspecific taxa. Due T. wiedersheimiisbasedon two specimens, a female from to the constant morphological differences between them Genderogebirge (=Tchabal Mbabo) and a subadult male andtheirallopatric distributionsweregardthe taxaquadri- from the village Tsch'a (Bekom), Bamenda area. In the cornis and eisentrauti as subspecies of a single species. course ofdescribing T. w.perreti, Klaver& Bohme (1992) Trioceros quadricornis quadricornis (Tornier, 1899) designated the female specimen as lectotype and conse- from the Manengouba area represents the nominate form quentlyrestricted the type locality to the Genderogebirge Bonn zoological Bulletin 57 (2): 211-229 ©ZFMK 218 Michael F. Barej et al. Fig. 3. Cameroonian chameleons (in life): A= Trioceros wiedersheimimale (Mt. Tchabal Gangdaba). B = T.perretimale (Kod- min, Bakossi Mts.)- C = T. serratus male (Kedjom Keku = Big Babanki, Bamenda Highlands). D = T. serratus female (Kedjom Keku= Big Babanki, Bamenda Highlands). E = Male neotype ofT. serratus (in alcohol; MNHN 2007.1494; Belo, Mt. Oku). F = Illustration of T. serratus after Mertens (1922; "Siidkamerun"). G = Male neotype ofT. serratus (in life; MNHN 2007.1494, Be- lo, Mt. Oku) in situ. Bonn zoological Bulletin 57 (2): 211-229 ©ZFMK Chameleons ofthe genus Trioceros from Cameroon 219 (=Tchabal Mbabo). As one of the uncovered molecular new material corresponded to the "paratypoid" (= clades contains specimens fromTchabal Mbabo, topotyp- paratype) ofT. serratus, while females are consistentwith ic material, thus this clade should correspond to T. wieder- the female cotype (=syntype) of T. wiedersheimi and fi- sheimi. The occurrence of T. wiedersheimi on Tchabal nally he recognizedthatthe prominence ofthe lateral and Gangdaba has already been assumed in the past (Chirio temporal cristae is subjectto individual variation. In a sub- & LeBreton 2007; Klaver & Bohme 1992) and we can sequent publication on material collected by Eisentraut at confirm its occurrence on this mountain range. Thereafter, Lake Oku and Lake Manengouba, Mertens (1968) con- the species is known from the northernmost parts of its firmed his fonner statement and remarked that males in former assumed distribution, while populations of the T. wiedersheimi also do possess a serrated dorsal crest, southern molecular clade from the Bamenda Highlands, while the dorsal crest is straight and simple in females. Mt. Mbam, and Mt. Oku represent a distinct taxon. It is With the exception ofLake Manengouba (recognized as alsovery likelythatthis clade covers populations fromthe distinctby Klaver& Bohme 1992), all localities ofthema- Mbulu Hills, Mt. Lefo and the Obudu Plateau in south- terial examined by Mertens belong to the newly discov- ernNigeriaas this would correspondto aborderingrange ered southern clade. from other studies (Gonwouo et al. 2006). Solely one lo- cality in direct proximity north ofthe Manengouba Mts. Ourmorphological analysis ofmaterial throughoutthe dis- (see map in Gonwouo et al. 2006) appears uncertain, as tribution range ofthe formertaxon T. w. wiedersheimire- T.perreti has been regarded as restricted to the mountain vealed that distinguishing characters chosen by Mertens range, but the specimen was not available to us for this (1922) are hard to assign to members of one clade, as study. many characters are present in members of both clades (lack ofheel spur, lack ofoccipital lobes, etc.) separating Mertens (1922) described Chamaeleon serratus from them from other taxa. Ofthe three main characters given "Siidkamerun" (= South Cameroon, Fig. 3F), being most by Mertens (1922), two ofthem seem to be inapplicable. similar to T. wiedersheimi, but differing by size, promi- According to Mertens (1922): (a) T. serratus grows larg- nence of the temporal cristae and course of the lateral er than T. wiedersheimi, but four often males from Tch- cristae (Fig. 3E). The species hasbeen later synonymized abal Mbabo and the Gotel Mts. (= T. wiedersheimi) pos- with T. wiedersheimi by Mertens himself(1940; see be- sess a larger body length than the largest member ofthe low). Klaver & Bohme (1992) argued that T. serratus is southern clade, and the largest female also belongs to T. a synonym ofT. wiedersheimi, as the original description, wiedersheimi; (b) temporal cristae are distinctin T. wieder- especially the low numberofscales on the scalloped dor- sheimi and indistinct in T. serratus, but this character sal ridge, isnot consistentwith T.perreti from Manengou- varies within bothclades (Fig. 4), whichwas alreadymen- ba Mts. A comparison with the type specimen of T. ser- tioned for Bamenda populations by Mertens (1940); and ratus was not possible, as the type specimen was proba- (c) Mertens (1922) differentiated the course ofthe later- bly destroyed duringthe SecondWorldWar(H. Wermuth al cristae (in front ofthe eye first running along the eye 16.4.1979 in lift., in Klaver & Bohme 1992). then in a weaker slope to the tip ofthe snout in T. wieder- sheimi, in contrast to an even slope in direction to the tip Mertens (1922) rightly suggested that T. wiedersheimi is of the snout in T. serratus) and this character is clearly morphologically the most similar species to T. serratus, more applicable to specimens belonging to the southern, but obviously he only compared his material with previously unrecognized clade (Fig. 4). Nonetheless, with Nieden's (1910) original description and notwith the type justa few specimens ofeach clade a determination on this specimen, as he only cited the original sections for com- character alone is difficult. At last, the name-giving char- parison. Beside the characters cited above, Mertens (1922) acter, a serrateddorsal crestis alsopresent in males ofboth mentioned that no additional distinct characters like the clades. The numberofscale rows forming the crenulation shape ofthe dorsal crest, which is atthe origin ofthe spe- is consistent in both clades (being formed ofup to three cific name (Mertens 1968), are given in Nieden's (1910) rows ofscales) andthe extent ofcrenulationalong thedor- description. [Remark: Nieden (1910) stated that a dorsal sum and base oftail also varies in both clades. Accord- crest is lackingbut, a dorsal midline is formedoftwo rows ing to Mertens (1922) each cusp of the crenulation is 3 mm mm of tubercle scales which are separated in groups of 3^1 highand4.5 long, butonly in oneverylarge spec- scales in the male specimen from the Bamenda region]. imen of T. wiedersheimi a comparable size has been reached, while specimens of similar size to Mertens' Later, Mertens (1940) reported on a collection delivered (1922) specimens possess smaller cusps in both clades. by M. Kohler including chameleons from the Bamenda The given type locality "Siidkamerun" does not allowany Highlands (fourmales + two females) and concludedthat direct localization ofT. serratus. Moreover, the subadult T. serratus is in fact ajunior synonym of T. wiedershei- male paralectotype ofT. wiedersheimioriginates fromthe mi. His conclusionwasbased onthe factthat males ofthe Bamenda area (part ofthe southern clade) and mighthave Bonn zoological Bulletin 57 (2): 211-229 ©ZFMK 220 Michael F. Barej et al. Fig. 4. Heads inlateral view. Row 1-3: Trioceroswiedersheimi(fromlefttoright): ZMB 21873 female(lectotype), ZFMK75744 female, ZFMK 68943 male, ZFMK 75740 male, ZMB 21857 male, ZMB 74805 female, ZFMK 78714 male, MNHN 2005.2753 male,NMP6V 74112 male, ZFMK 75745 female, MHNG 1544.001 male, MHNG 1544.002 male. Row4-6: T. serratus (from left toright): MNHN2007.1494male(neotype), MNHN2007.1465 female,NMP6V74104male, MHNG 1365.023 female,ZSM 13.1925 MNHN ZFMK MNHN MNHN MHNG subadultmale, 2007.1464male, 5801 male, 2007.1463 male, 2007.1461 female, 1365.024 MHNG ZFMK female, 1010.049 male, 5800 female. Bonn zoological Bulletin 57 (2): 211-229 ©ZFMK