Chromaphyosemion Radda

Rainer Sonnenberg, 2007, Description of three new species of the genus Chromaphyosemion Radda, 1971 (Cyprinodontiformes: Nothobranchiidae) from the coastal plains of Cameroon with a preliminary review of the Chromaphyosemion splendopleure complex., Zootaxa 1591, pp. 1-38 : 20-28

[[ Genus Chromaphyosemion Radda ]]

DNA analysis

The alignment contains 795 bp for a set of 19 specimens and 713 bp for C. alpha . There were 238 variable positions of which 155 are parsimoniously informative. The average amount of uncorrected pair-wise sequence divergence was 8.08%, the maximum observed value 18.82% between C. alpha and C. riggenbachi . The sequences were translated into 264 amino acid positions of which 21 were variable and eight were parsimoniously informative. Average nucelotide frequencies were A = 24.9 %, C = 29.2 %, G = 15.7 %, and T(U) = 30.3 %, showing the characteristic anti-G bias of the mitochondrial genome base composition ( Zhang & Hewitt 1996).

The MP (Fig. 15) and Bayesian analyses (Fig. 16) yielded nearly identical topologies. Numbers on branches represent bootstrap values (MP topology) or posterior probabilities (Bayesian topology). The Bayesian tree shows more polytomies but does not contradict the MP tree. Chromaphyosemion alpha is the phylogenetically basal species within Chromaphyosemion according to a larger dataset with different outgroup species (unpubl. data and Agnèse et al. [2006]). The newly described C. melinoeides is the sister species to C. lugens . One well supported clade in both trees contains, besides C. loennbergii and C. punctulatum , two equally well supported groups of closely related species, marked as NC and SC. NC is the northern clade, a group which has its distribution approximately north of the Wouri. SC is the southern clade which is found south of the Wouri (Map 1). Within the northern clade, C. poliaki , C. splendopleure and C. volcanum have been described by earlier authors. The southern clade contains, besides the undescribed populations from Bioko and Bimbia, the newly described species C. koungueense and C. omega . Both had been formerly included in C. splendopleure ( Agnèse et al. 2006; Amiet 1987; Huber 1998a; Legros & Zentz 2007; Wildekamp1993). Within both groups genetic differentiation is low.

The mitochondrial cytochrome b sequences and the corresponding diagnostic region of a nuclear LSU fragment were compared for a subset of specimens, including two additional samples from populations with relevance for conflicting taxonomic hypotheses. This comparison shows incongruence in the grouping of C. splendopleure collected near Tiko (the presumed type locality) and the population of Mboro (CMM 18, as C. splendopleure in Agnese et al. [2006] ) (Figs. 17 and 18). The nuclear marker does not support the inclusion of these populations in the southern clade as indicated by the mitochondrial fragment. Instead, the C. splendopleure from the area of Tiko is grouped into the northern clade ( C. poliaki and C. volcanum sensu Agnese et al. [2006] ), and the population from Mboro into C. punctulatum , which in both cases is congruent with diagnostic color pattern characters.

Discussion

Research on the genus Chromaphyosemion was hampered until recently by a missing phylogenetic hypothesis and some confusion caused by insufficiently diagnosed and delimited taxa. Earlier attempts based on morphology, crossing experiments, karyotypes and color patterns could not resolve the relationships within this genus ( Huber 1998a; Scheel 1966, 1968, 1974, 1990; Sonnenberg 2000). In his last publication, Scheel (1990) underestimated the diversity of Chromaphyosemion as now presented by the molecular phylogenies (Figs. 15- 16, Agnèse et al. 2006; Legros & Zentz 2007). Especially C. bitaeniatum in his sense is clearly a mixture of several different lineages with respect to the cladograms in Figures 15-16. It includes C. alpha , C. bitaeniatum , C. kouamense , C. koungueense , C. melanogaster , C. omega , C. poliaki , C. punctulatum , and C. splendopleure .

The value of color patterns as phylogenetic characters besides their doubtless taxonomic utility is very limited, and the grouping of Chromaphyosemion species given by Huber (1998a) is only partially compatible with the currently published molecular phylogenies. His groups turn out as para- or polyphyletic in the molecular phylogenies (Figs. 15-16, Agnèse et al. 2006; Legros & Zentz 2007). Several characters used in his assignments intergrade: e.g., the absence of red or dark markings on anal fin in the C. bitaeniatum subgroup and the more or less streaked or spotted anal fin in the C. loennbergii subgroup. Streaks or dots in caudal and dorsal fins have no or only limited phylogenetic value because they occur in all groups and their presence can sometimes even vary within one species (e.g., C. loennbergii , Figs. 19-20). In addition red dots, red reticulation, or red blotches on sides are basically the same character and differ only quantitatively. Huber's (1998a) C. bitaeniatum group is similar in its composition with Scheel's species C. bitaeniatum , and thus also not monophyletic (Figs. 15-16). It contains, besides the basal C. alpha , all species of the northern and southern clades and C. punctulatum , but not C. loennbergii and C. poliaki . The inclusion of C. poliaki into his third group with C. bivittatum and C. lugens is not supported by the molecular phylogeny. In general, this subdivision of Chromaphyosemion conflicts in several points with the published molecular phylogenies ( Agnèse et al. 2006; Legros & Zentz 2007).

The cladograms presented here are compatible with recently published results ( Agnèse et al. 2006; Legros et al. 2005; Legros & Zentz 2007). In cases of incongruences, these involve nodes with no significant bootstrap support or posterior probability and might be caused by the different mutation rates of the considered mitochondrial genes or unsufficient information content of the fragments.

The analysis of the mitochondrial DNA data including all described species currently accepted as valid and most undescribed forms gives an overview of the relationships of this diverse genus. It allows the comparison of color patterns between the most closely related species and the evaluation of the distribution of certain patterns within a phylogenetic framework. Whereas earlier authors were overwhelmed by the diversity of male coloration and had mostly only sparse distribution data, it is now possible to recognize patterns in the distribution of the different color forms. Most of these patterns appear to be geographically localized groups of populations that correspond to already-named species. They are distinct from their surrounding congeners and represent well-separated units and are, with regard to the molecular data ( Agnèse et al. 2006; data presented here), well differentiated, supporting the value of color pattern characters in Chromaphyosemion taxonomy.

Chromaphyosemion melinoeides has a more basal position in the phylogenies (Figs. 15-16) and is distinguished from its closest relative, C. lugens , by consistent male and female diagnostic color patterns, which are stable over the known distribution area. It is assumed that the differences are not within the normal variation of these species and they are regarded as distinct evolutionary units, separated by stable, diagnostic characters which most probably belong to a complex of characters employed in mate-choice and/or species recognition. In addition, the currently known karyotypes of the two species differ considerably, with diploid chromosome numbers of 2n=36 in C. lugens and 2n=28 in C. melinoeides ( Vôlker et al. 2005; M. Vôlker pers. comm.). The different karyotypes might have an influence on reproductive isolation between the two species. However, as mentioned before, karyotype variation is common in Chromaphyosemion ( Vôlker et al. 2005, 2006, 2007a,b, in press; Scheel 1974), but the impact of karyotypic differences on reproductive incompatibilities is not known in detail and difficult to prove ( King 1993; Rieseberg 2001).

However, within the most recent radiations (NC and SC, Figs. 15-16) molecular and karyotypic differentiation is weak despite pronounced differences of male color pattern between the included species (see e.g., C. poliaki , Fig. 21 and C. splendopleure , Figs. 22-23) ( Agnèse et al. 2006; Völker et al. 2007b, in press; Völker pers. comm.). This is in correspondence with the observation that characters under sexual selection can evolve at higher speed than others (e.g., Pomiankowski & Iwasa 1998; Uy & Borgia 2000), which is a possible explanation for the rapid speciation of cichlids in the great African lakes ( Maan et al. 2004; Salzburger et al. 2006; Seehausen 1999).

Available names for population groups within the northern and southern clades were C. splendopleure , C. volcanum and C. poliaki . Whereas C. poliaki , which is endemic to the southern and southeastern flanks of Mount Cameroon, is easily distinguished from all surrounding populations by male color pattern (Fig. 21), it is difficult to find diagnostic and constant differences between the Chromaphyosemion populations from Tiko and Kumba, the areas around the type localities of C. splendopleure (Fig. 22) and C. volcanum (see Eberl 1996, p.52 for a picture of a specimen from the type locality), respectively. Chromaphyosemion volcanum was therefore until recently regarded as a synonym of C. splendopleure ( Legros 1991; Sonnenberg 2000). The mitochondrial DNA analyses groups populations which are determined as C. splendopleure and C. poliaki based on coloration characters, together with some unnamed forms, into the northern clade. Additionally the distinct status of some of the populations of the southern clade was often recognized by earlier authors ( Eberl 1996; Eigelshofen 1998; Legros et al. 2005; Sonnenberg 2000) but until now without taxonomic consequences.

After submission of this manuscript, Agnèse et al. (2006) published a molecular study of Chromaphyosemion including a population from the vicinity of the proposed type locality of C. splendopleure, Tiko in Cameroon. Whereas the resulting phylogeny is largely congruent with that presented here, some of his taxonomic conclusions are different: the mitochondrial sequences of the Tiko population (CBL 01/25, entrance of Tiko) cluster within what is here named southern clade (Figs. 15-16). It includes the newly described species C. koungueense (Figs. 3-5), C. omega (Figs. 9-10) and the northernmost populations of C. punctulatum (Fig. 24), which are all clearly distinguished from C. splendopleure (Figs. 22-23) by color pattern. Agnèse et al. (2006) assigned the name C. splendopleure to the Tiko population and the species mentioned in the previous sentence, based on mitochondrial DNA cluster, and state that this taxon 'exhibits the highest degree of phenotypic variability' and transferred the populations of the northern clade to C. poliaki and C. volcanum . This taxonomic decision based on mitochondrial DNA did not take into account the diagnostic coloration characters and the incongruence was regarded as high phenotypic variability without further discussion and was followed in a later paper by Legros and Zentz (2007).

In order to clarify the incongruences, I sequenced a variable part of the nuclear large ribosomal subunit (LSU) to study the discrepancies with an additional independent marker. The comparison of the mitochondrial clusters and an excerption of the alignment of the nuclear gene fragment showing the most informative region (Figs. 17-18) indicates incongruences in two cases that are important for taxonomic conclusions: a) the clustering of the C. splendopleure collected near Tiko within the southern clade (SC) by the mitochondrial data is not congruent with the presented diagnostic LSU sequence fragment and b) a similar incongruence of the Mboro population. In both cases the grouping by diagnostic coloration characters is compatible with the LSU fragment but not with the clustering of the mitochondrial DNA. The inclusion of C. koungueense , C. omega and the northern populations of C. punctulatum in a 'polymorphic' taxon C. splendopleure based on mitochondrial DNA cluster is not supported by the nuclear and coloration pattern data; therefore, the taxonomic conclusions of Agnèse et al. (2006) and Legros and Zentz (2007) are not followed here. The populations around Tiko belong, according to nuclear DNA and coloration characters, to the northern clade and C. volcanum is again regarded as a synonym of C. splendopleure . The southern clade includes C. koungueense and C. omega , together with the populations from Bioko and Bimbia. The population from Mboro is regarded as a belonging to C. punctulatum , according to nuclear DNA and diagnostic color patterns.

The results of the nuclear marker are congruent with the species assignment by diagnostic coloration characters in those cases where the mtDNA gives a different result. As also coloration pattern characters have to be regarded as nuclear encoded marker, the observation of incongruences with the results of the mtDNA study is in line with a most probable mitochondrial introgression in the studied populations. The problem of species assignment based on only one marker, especially a mitochondrial marker, is currently discussed in the frame of genetic 'barcoding' approaches (e.g. Moritz & Cicero 2004; Sonnenberg et al. 2007).

This is important in so far as Agnèse et al. (2006) use only the mitochondrial DNA cluster as argument for their inclusion of the populations of the southern clade (SC) into C. splendopleure , which then includes the C. splendopleure population from the Tiko area and the here described species C. koungueense and C. omega in addition to the northern populations of C. punctulatum which can all be assign by coloration pattern of males to different groups. As the diagnosis by coloration and the results of the LSU sequences are compatible, this means that at least two independent marker (if coloration is counted as only one nuclear encoded marker) are congruent and differ in the same cases with the mitochondrial results, it seems most probable that the conflicting mitochondrial data are caused by mitochondrial introgression.

Despite these results, the largest taxonomical problem within this genus was and in some part is the wastebasket taxon, C. splendopleure . This species was described from a commercial import in 1929 ( Brüning 1929a), collected around Tiko at the foothills of Mount Cameroon. It was subsequently often synonymized with C. bitaeniatum or C. bivittatum ( Scheel 1966, 1968, 1974, 1990) or sometimes includes nearly all Chromaphyosemion found on the Atlantic side of the coastal plains, except for C. bitaeniatum ( Amiet 1987; Huber 1998a, b; Radda & Pürzl 1982, 1987; Sonnenberg 2000). In recent years several new species were described from this taxonomic wastebasket, namely C. alpha , C. koungueense , C. melanogaster , C. omega , C. poliaki , and C. punctulatum ( Amiet 1991; Huber 1998a; Legros et al. 2005; this paper). All these species are only more or less closely related and their inclusion in C. splendopleure was never based on diagnostic characters. The type locality is only approximately known and within this area several differently colored forms can be found. Currently a form with very red to orange coloration on fins and body can be distinguished (Figs. 22- 23, as SPP in Figs. 15-18, Map 1), which is found around Tiko, Moliwe and also in the area of Kumba (which was described as C. volcanum ) and shares many characters with C. splendopleure as given in the description by Brüning (1929a) and later Meinken (1930a); a second form with bluish to pink sides and yellow to greenish fins (Fig. 25, as cf SPP in Figs. 15-18, Map 1), which is known from the eastern flanks of Mount Cameroon at Owe, Lykoko and Muyuka and further north with one population known from Oron (Cross river system) in Nigeria ( Radda & Pürzl 1982; Sonnenberg 2000); and a group of populations called C. sp. 08 which were first mentioned by Amiet (1987) and considered to be C. loennbergii populations north of the Sanaga (Fig. 26, as sp8 in Figs. 15-18; Map 1). Additionally there are several populations found southeast of a line Tiko-Kumba, which are difficult to add to one of the former three forms based on color patterns. All these populations, together with C. poliaki (Fig. 21), form a weakly differentiated group according to the mitochondrial data ( Agnèse et al. 2006, Figs. 15-16). The exclusion and description of C. koungueense and C. omega in this publication restricts the still problematic taxon C. splendopleure to a smaller and phylogenetically closely related groups of populations, which form the northern clade. Still it is currently difficult, with the exception of C. poliaki , to give unambiguous combinations of diagnostic color characters for these forms. In addition, it remains problematic to assign the specimens, on which the description of C. splendopleure is based, to one of two biogeographically possible forms. As the description of C. splendopleure was based on live specimens, a lot of information on male color pattern is known ( Brüning 1929a, b; Meinken1930a) but actually it is not possible to decide on which form the description is based. I currently find better evidence for the more reddish form with orange-yellow anal fin, based on Brünings (1929a) and Meinkens (1930a) color description (sides metallic golden or greenish-blue, ventral and on caudal peduncle golden yellow, color of dorsal fin red-brown in center, anal fin greenish-yellow at base and golden-yellow in distal part ( Brüning 1929a; Meinken 1930a)). This opinion is expressed in the application of the name C. splendopleure for this form in the present paper. It does not anticipate any other decision in the future and a final solution has to be found before the taxonomy of this group can be solved.

As a first step toward a stable taxonomy, which takes phylogenetic information into consideration, it is now possible to remove all the taxa included in C. splendopleure sensu Huber (2000) or C. bitaeniatum sensu Scheel (1990) , which are not related to the clade containing C. splendopleure s.s. (the northern clade NC) and are distinguishable by stable characters of the male color pattern. This allows the reduction of a polymorphic or variable species into diagnosable units with the additional argument of support of phylogenetic information. I want to stress that the diagnoses of the new species are also possible without the molecular phylogeny.