A New Species of Cyprinodont fish, Aphyosemion tirbaki n. sp. (Cyprinodontiformes: Aplocheilidae) from Gabon, with further Evidence of the Frontier Species Concept.




Abstract: Aphyosemion tirbaki n. sp. is described from two adjacent localities in the primary rainforest of southeastern Gabon. A member of the Aphyosemion ogoense superspecies complex, it is distinguished from other members of this complex by a combination of morphometric and meristic features, as well as male color pattern. Males bear a striking resemblance in color to Aphyosemion coeleste, a vicariant species belonging to a different species group. These species provide an example of the "frontier species" concept, which is defined as a cryptic species of one group occurring, in a restricted area, near the periphery of the geographical range of the group, and living in close geographic proximity to a congener belonging to a different species group, possibly with a color convergence of both species.


Résumé: Aphyosemion tirbaki n. sp. est décrit de deux localités voisines dans la forêt primaire du Gabon sud-oriental; il est séparé des formes phylogénétiquement proches appartenant à la superespèce A. ogoense par quelques caractères morphoméristiques et par la coloration du mâle; en revanche, le mâle montre une étonnante convergence de coloration avec A. coeleste, une forme vicariante appartenant à une superespèce distincte: c'est un nouveau cas, chez Aphyosemion, d'espèce frontière, un concept qui est défini comme l'apparition d'une espèce cryptique, habitant une aire restreinte à la limite de la distribution de sa superespèce, faisant face à une autre espèce cryptique d'une autre superespèce, éventuellement avec convergence de coloration des deux formes.



The genus Aphyosemion Myers, 1924 is a group of tropical fishes characteristic of small, quiet brooks in the primary rainforests of western Africa. Species are found living either singly or more often sympatrically in groups of up to five congeners, although in cases of sympatry the respective ecological niches are not obvious (Huber, 1978). The often spectacular colors of the males have resulted in a high degree of popularity among aquarists specializing in Cyprinodonts (oviparous Cyprinodontiformes). This in turn has led to a number of taxonomic and karyological studies, as well as increased field investigations in Gabon and adjacent areas of western Africa inhabited by these fishes.

The history of collections in southeastern Gabon illustrates well the situation.

For many years the genus Aphyosemion was known from that area only from a single species, which was first collected from the vicinity of Franceville during the late 1920's by the French civil administrator, Msr. Baudon and which was described by Pellegrin (1930) as Haplochilus lujae var. ogoensis.

No further ichthyological investigations of this area occurred until 1976, at which time a specialized expedition comprised of Radda, from Vienna University, Austria, and the author made collections that resulted in descriptions of the following four new species of Aphyosemion: A. lamberti Radda and Huber, 1977; A. rectogoense Radda and Huber, 1977; A. coeleste Huber and Radda, 1977; and A. punctatum Radda and Pürzl, 1977. Since then, more than a dozen expeditions to that region have resulted in the discovery of additional populations of the above taxa, as well as three additional species that were first described from the adjacent Republic of the Congo (formerly French Congo): A. schluppi Radda and Huber, 1978; A. pyrophore Huber and Radda, 1979; and A. buytaerti Radda and Huber, 1978.

The small area where the above species are collected, which encompasses only about 1500 square kilometers, is part of a homogeneous plateau known as the Massif Du-Chaillu, in central and southern Gabon and neighboring parts of Congo. There occurs a remarkable diversity of Cyprinodonts, including over twenty species of Aphyosemion, one of the genus Epiplatys, one of the genus Procatopus and two of the genus Hypsopanchax. Although fish have so far been collected at over a hundred sites (Huber, 1996), these have been limited to the southern half of the region, which is the only area accessible by road. Considering the high diversity of killifishes therein, it is likely that additional species remain to be discovered in the still-unexplored northern areas.

The new nonannual species of Aphyosemion alluded to above, A. tirbaki, is described on the basis of material collected in 1996 from Gabon. It belongs to the subgenus Mesoaphyosemion, which comprises 51 species and is the largest of the 12 subgenera of Aphyosemion listed by Huber (1996). A. tirbaki is a member of the A. ogoense species group (i.e., superspecies), which is considered to contain nine species. As is true of other superspecies, the ogoense group is characterized by similarities in pigmentation pattern, life colors, and morphometric and meristic features, with the included forms differing in details of male pigmentation and color, as well as karyology. These are discussed and illustrated in the present paper. The species groups of Aphyosemion found in that part of the Massif Du-Chaillu plateau are also discussed in this paper. Aphyosemion tirbaki bears a striking resemblance to A. coeleste, a vicariant species belonging to a different species group within the subgenus Mesoaphyosemion. Although superficially similar, A. tirbaki and A. coeleste differ in a number of meristic and morphometric features, which are compared later in this paper. They are examples of "frontier species" which are defined as cryptic species of a group occurring near the geographical periphery of that group. Such a species in turn characteristically lives in close geographic proximity to a congener belonging to a different species group, possibly with color convergence between the two species (Huber, 1977). The "frontier species" concept is discussed in the present paper.

Methods.- Counts and measurements were made according to the methods originally described by Hubbs and Lagler (1958) and modified by Huber (1992), for the related neotropical Cyprinodont genus Rivulus. Measurements were made with dial calipers, read to the nearest 0.1 mm and the percentages, in standard length (S.L.), were rounded, except for the computation of the mean values and the standard deviations. Radiographs were made of a small sample (eight specimens) in order to verify fin-ray counting procedure, and especially to verify previous estimates made (using a binocular microscope) of the relative positions of the dorsal and anal fins (termed the D/A ratio). A D/A value of +4 indicates that the dorsal-fin origin is situated above the fourth anal ray, a +5 value signifies a position above the fifth ray, etc. The frontal squamation (scalation) pattern follows Hoedeman (1958) and the cephalic neuromast pattern follows Clausen (1967). Both methods are also detailed in Huber (1978). Institutional abbreviations are as listed in Leviton et al. (1985).

Genus Aphyosemion Myers, 1924.-

Myers (1924) described the genus Aphyosemion with the following characters, diagnostic and opposite to those of the Indo-asian and African genus Panchax Valenciennes (= Aplocheilus Mac Clelland, including Epiplatys Gill, in their present definition). They still stand firm (Huber, 1978): a less flat dorsal profile of head; a snout, shorter than eye diameter;. a lyre-shaped caudal fin. In the same paper, he also described Fundulopanchax, as a subgenus of Aphyosemion, distinguished by "the dorsal fin being inserted forward, above or very slightly behind the anal origin, and the subequality of these two fins".

Parenti (1981: 476-479) added several osteological characters to the definition of both Aphyosemion and Fundulopanchax and upgraded the latter to the genus level. The former was diagnosed by "being nonannual, having a dorsal fin count of 7 to 14 rays, which is situated no farther anteriorily than opposite the first quarter of the anal fin origin, and possessing a swimbladder extending posteriorily to the first one or two haemal arches". The latter, in contrast, is an annual, with over 14 rays and dorsal and anal fins near to opposite, with a non-extending swimbladder. However, her view was not followed by several subsequent authors (e.g. Wildekamp, Romand & Scheel, 1986; Romand, 1992; Huber, 1996), with the following arguments: first, the distinction between the annual and nonannual conditions is not clear cut, with a continuum of states and semi-annual species as intermediates; second, some nonannual subgenera (e.g. Archiaphyosemion Radda, 1977, then available and Scriptaphyosemion Radda and Pürzl, described later, in 1987) were thought to be phylogenetically related to an annual subgenus Callopanchax Myers, 1933, because all share a common micromorphological character (a closed pattern of the first frontal pair of neuromasts, as per Clausen, 1967); third, strictly nonannual subgenera, like Chromaphyosemion Radda, 1971 and Diapteron, Huber & Seegers, 1977, placed by Parenti in Aphyosemion s.s. show opposite and subequal dorsal and anal fins (but, with less than 14 rays at dorsal); last, annual or semiannual subgenera, like respectively Paludopanchax Radda, 1977 and Paraphyosemion Kottelat, 1976, placed by Parenti in Fundulopanchax, show less than 14 dorsal fin rays in some species.

Preliminary recent studies, though, which are based on mitochondrial genes (Murphy & Collier, 1997) or on otholiths (Zee & Wildekamp, 1995), suggest that Aphyosemion sensu Myers is paraphyletic. It may be further subdivided in the future, but not accordingly to Parenti's delineation and diagnosis.

The new species, tirbaki, falls anyhow more in Aphyosemion than in Fundulopanchax, sensu Parenti: it is a strict nonannual, the dorsal and the anal fins are not subequal and not opposite, the swimbladder is extended. However, the dorsal fin count is at the upper limit (14-15) and the D/A ratio is unusually low (+3 or +4, near to the first quarter of the anal fin).

Subgenus Mesoaphyosemion Radda, 1977.-

Radda (1977) described 4 new subgenera of Aphyosemion on the basis of meristics, of biological conditions (annualism or not), of karyological data, on key characteristics of color pattern and on distribution. For example, Mesoaphyosemion was described with 9 to 13 dorsal fin rays, 13 to 17 anal fin rays, 28 to 34 scales in lateral line, a D/A deviation of +5 to + 10, and by encompassing nonannual species. The distribution is restricted to the central west African forest. No comparative diagnosis was provided for these subgenera and the other 7 already described subgenera and overlapping characters were part of the definitions. A twelfth subgenus, Scriptaphyosemion was subsequently described by him, together with Pürzl in 1987, with the same approach and limitations.

However, Radda assigned each so far known species to a subgenus. The new species, A. tirbaki, falls readily in the subgenus Mesoaphyosemion, because its closest relatives, the members of the ogoense superspecies (see further, the superspecies concept) are all listed in Mesoaphyosemion, because of its distribution, and because of its meristics. The meristics do not fit, though, entirely with Radda's description.

This is not the scope of this study to present a review of these 12 subgenera (Parenti simply listed them, either in Aphyosemion or in Fundulopanchax), or to determine if Mesoaphyosemion is a valid subgenus, all the more that several other researchers are in the process of doing so, separately (e.g. Costa, Collier, Zee; pers. comm.). And the subgenus status will not be discussed further in this paper.

The Superspecies Concept.-

The genus Aphyosemion may be divided into a number of species groups, which are characterized by identical, or at least very similar, meristic and morphometric characters (isomorphism), and often are distinguished with certainty only by differences in male color pattern and by karyological differences. Scheel (1968) was the first to analyze these species groups, and his results were later expanded and elaborated upon by me (Huber, 1978).

Superspecies are known (Huber, 1992) for the related neotropical genus Rivulus, which ranges from southern North America south into South America, including the West Indies. Validity of these superspecies has been confirmed by the study of Murphy and Collier (1996), who expanded previously-held concepts by providing evidence that these monophyletic groups may also include certain aberrant non-isomorphic species.

The Aphyosemion species dwelling the region of the biogeographically homogeneous Massif Du-Chaillu plateau, where tirbaki is collected, can be grouped into five superspecies, based on key external characters. First, the ogoense superspecies, is characterized by a medium size (total maximum length in aquarium: 50-70 mm), by a higher number of dorsal fin rays (D= 11-14, rarely 15), a lower D/A deviation ratio (D/A= +3 to +6) and by a lack of streamers in the unpaired fins. In the region, species include: ogoense, pyrophore. Second, the coeleste superspecies, also with rounded fins, is distinguished by a smaller size (T.L.: 40-50 mm), a slightly less deep body than the ogoense superspecies (19-20% versus 21-22% in S.L.), by a lower number of dorsal rays (D= 10-11), by a complementary higher D/A deviation (D/A= +6 to +8) and by a post-opercular golden-pink contrasting zone with or without red markings. In this region, species include: coeleste, passaroi. Third, the wildekampi superspecies is only distinguished from the coeleste superspecies by the spade shape of the caudal fin, the pointed dorsal fin and the lack of the post-opercular characteristic zone. In this region, species include: punctatum, aureum. Fourth, the elegans superspecies is distinguished from the 3 above superspecies by a slenderer body (depth= 17-18% in S.L.), by a short remote dorsal fin (D= 7-9; D/A= +8 to +11), and by the presence of medium sized streamers on unpaired fins of male. Its size is similar (T.L.= 40-50 mm) to the coeleste and wildekampi superspecies. In this region, species include: lamberti, rectogoense. Fifth, the batesii superspecies is distinguished from all the above by the larger size (T.L., more than 80 mm, up to 150 mm), by the higher scale counts (LL= 32-35 versus 27-31 for the 4 above superspecies), by the higher and subequal counts of unpaired fins rays (D= 15-17; A= 15-17), by long streamers on male unpaired fins and by its annual condition. In this region, species include: splendidum.

The ogoense superspecies is the most widespread and the most species rich in the Massif Du-Chaillu. It encompasses eight member species which are the following, with the corresponding respective meristics on average of limited material (Huber, 1996): A. ogoense (D= 11; A= 14; D/A= +5; LL= 30), A. ottogartneri (11, 15, +6, 30), A. pyrophore (12, 15, +6, 30), A. caudofasciatum (11, 14, +5, 30), A. louessense (13, 15, +5, 31), A. zygaima (12, 15, +6, 31), A. labarrei (13, 15, +6, 31), A. joergenscheeli (12, 15, +4, 29).

Based on its morphological characters, we infer that the new species A. tirbaki is a component of the ogoense superspecies (maximum T.L.: 70 mm; rounded unpaired fins; D= 14-15; D/A= +3). However, as stated in the introduction, the Cyprinodont fauna of the northern Du-Chaillu plateau is unprospected and this hypothesis may be reappraised following future collections. A. tirbaki, as a component of the ogoense superspecies, is best separated from the other components by the male color pattern, made of a series of broad irregular red markings on the sides and of a symmetrical organization of external bands on the caudal fin, comprising a yellow margin and a red submargin, whereas the other components are differently colored (figure 1). The ogoense s.s. male shows a series of red longitudinal spots on the sides including a broad basal red line from the anal insertion to the lower caudal and an asymetrical organization of external bands on the caudal fin (red margin and yellow/blue submargin on the upper part; the reverse on the lower part), plus a flamed pattern in the center of the fin. The ottogartneri male is similar to ogoense, but without the typical basal line on the sides. The pyrophore male is similar to ottogartneri, but the red longitudinal lines on the sides are replaced by thin red bars, posteriorily. The caudofasciatum male is similar to pyrophore, but with a red fasciated, instead of a flamed caudal fin. The louessense male is similar to ogoense, but the series of longitudinal spots are replaced by variable irregular blotches and a single broad mid-line on the sides. The zygaima male is similar to louessense, without the mid-line, but the caudal organization of external bands is symmetrical. The labarrei male is similar to zygaima but the irregular spots are replaced by vertical blotches and the caudal fin is lacking the yellow submargins and the margins are black, not red. The joergenscheeli male is vaguely similar to ottogartneri, but the series of side spots are arranged rather vertically and the caudal pattern is unique with a red circumcaudal submargin on a plain deep blue background.

All taxa for which the karyotype has been studied are determined by separate fundamental numbers and breed true to their characteristics, which discards the possibility of simple color variations or natural hybridization.

The inclusion of tirbaki with its symmetrical caudal pattern like caudofasciatum, zygaima, labarrei and joergenscheeli allows us to notice that all symmetrically patterned species are distributed with a small range at the periphery of the superspecies (see further, distribution). All nine taxa represent a monophyletic line of distinctive cryptic biospecies, replacing each other vicariantly, the central phenotype, largely distributed, named louessense, displaying the most variable and generalized pattern (Huber, 1980), like in many Aphyosemion superspecies (e.g. the cameronense or gardneri superspecies).

Aphyosemion tirbaki , new species

Figure 2, live male and figure 3, live male of A. coeleste.

Holotype.- MNHN 1997-186, a male, 38.5 mm S.L. and 46.85 mm T.L., from a forest "marigot" near Tsotandzala, a village on the road from Lastourville to Moanda, 82.7 km from Lastourville, or about 38 km from Moanda, Joumini River, a part of Youmidié, then of Lékoudi basin (geographical coordinates in degrees and hundredths of degree: 1.28S; 13.03E); locality code: EBT 96/7; altitude: about 700 meters. W. Eberl, T. Blum and P. Tirbak, Coll. March 3. 1996.

Paratypes.- MNHN 1997-187, 18 males; MNHN 1997-188, 26 females; NMW-93301, 2 males and 2 females; CAS 96952, 4 males and 4 females; BMNH-1997.8.28.7-14, 4 males and 4 females; MRAC-97-44-P-7-10, 4 males and 97-44-P-11-14, 4 females; UFRJ 3730, 4 males and 4 females; NRM 37494, 4 males and 4 females; ANSP 176179, 4 males and 4 females; all paratypes have been collected by the same collectors, at the same time and place than the holotype (paratopotypes).

Diagnosis.- A. tirbaki is distinguished from all species of Aphyosemion sensu Myers by the combination of the following six characters. First, the relative position of the dorsal fin insertion, not far, vertically, from that of the anal (D/A= +3 or +4, i.e. 3 or 4 rays behind the anal fin insertion): this is a small deviation in Aphyosemion and the smallest one among the non-annual Aphyosemion species from the Southern Gabon and Congo plateau. Second, the high number of dorsal fin rays (D= 14-15): this is the highest record in the genus if the annual species are excluded. Dorsal counts of non annual species vary on average from D= 8 to D= 13 (Huber, 1996). Third, the somewhat deeper body at the anal fin level (depth: 21% in S.L.) and at the peduncle level (resp. 12-13%). These measurements are lower in the other non annual Aphyosemion species of the region, except for the ogoense superspecies: the same proportions vary respectively from 17% to 20% and from 9% to 11% for them, on preserved field specimens (not aquarium bred specimens which get fatter). Fourth, the male color pattern, made of a broad yellow margin and a red submargin on vertical fins and irregular red vertical markings on sides, without any post-opercular contrasting zone. Fifth, the rounded vertical fins, without extended rays. Sixth, the non-annual ecological conditions, which implies a shorter incubation period of the eggs, in water: annual species habitats completely suffer from desiccation during the dry season and the eggs undergo compulsory diapauses.

Among the over-120 presumably valid species of the genus (Huber, 1996) none is exhibiting those combined characters: only one species, A. gardneri s.l. from the coastal lowlands and corresponding foothills of Nigeria and northwestern Cameroun is coming close to this combination; however, it differs from the new species by its annual or semi-annual condition, by the presence of short streamers on the male caudal fin, by an even deeper body (depth at anal level: 22%-25% of S.L., according to Scheel, 1968). We infer that those shared characteristics of these two distantly related species are just the result of hazard, the total number of combinations being in essence limited.

Description.- The key morphomeristical characters of eight types are the following (the holotype, first; average and standard deviation, last).

Sex: male, male, male, male, male, female, female, female. Standard length (in mm): 38.5; 36.8; 36.6; 37.9; 34.0; 30.5; 29.5; 32.3. Total length (in % of S.L.): 122%, 124%, 125%, 122%, 121%, 123%, 125%, 124% (mean: 123.25%; S.D.: 1.35%). Predorsal length (in % of S.L.): 62%, 64%, 65%, 63%, 64%, 65%, 67%, 65% (mean: 64.50%; S.D.: 1.58%). Preanal length (in % of S.L.): 60%, 61%, 61%, 58%, 60%, 63%, 63%, 60% (mean: 60.71%; S.D.: 1.72%). Preventral length (in % of S.L.): 47%, 53%, 50%, 48%, 49%, 51%, 49%, 48% (mean: 49.54%; S.D.: 1.84%). Depth at Anal level (in % of S.L.): 21%, 20%, 22%, 21%, 21%, 21%, 22%, 21% (mean: 21.10%; S.D.: 0.62%). Depth at Peduncle level (in % of S.L.): 12%, 13%, 13%, 13%, 13%, 12%, 13%, 13% (mean: 12.50%; S.D.: 0.48%). Head length (in % of S.L.): 26%, 30%, 27%, 28%, 27%, 29%, 27%, 24% (mean: 27.14%; S.D.: 1.73%). Interorbitar (in % of S.L.): 16%, 13%, 13%, 12%, 13%, 14%, 13%, 12% (mean: 13.18%; S.D.: 1.23%). Eye diameter (in % of S.L.): 8%, 8%, 8%, 7%, 7%, 9%, 8%, 7% (mean: 7.52%; S.D.: 0.56%). Snout (in % of S.L.): 7%, 8%, 7%, 6%, 7%, 8%, 7%, 7% (mean: 7.16%; S.D.: 0.56%). Vertebrae (abdominal+caudal): 12+16, 12+15, 13+15, 12+16, 11+16, 12+16, 12+15, 11+16 (mean: 27.5; S.D.: 0.5). Dorsal fin rays: 15, 15, 15, 15, 14, 14, 14, 14 (mean: 14.5; S.D.: 0.5). Anal fin rays: 17, 17, 16, 16, 16, 16, 16, 16 (mean: 16.25; S.D.: 0.43) D/A deviation: +4, +4, +3, +4, +3, +3, +4, +3 (mean: +3.5; S.D.: 0.5) LL (lateral scales + scales on caudal peduncle): 28+3, 27+3, 26+4, 28+3, 28+3, 27+3, 28+2, 27+2 (mean: 27.4; S.D.: 0.7). pDor (predorsal scales): 16, 16, 17, 15, 15, 16, 15, 16 (mean: 15.8; S.D.: 0.7). TRAV. (transversal scales): 9, 8, 8, 8, 8, 8, 8, 8 (mean: 8.1; S.D.: 0.3). CIR (circumpeduncular scales in a transversal zig-zag row): 16, 16, 15, 14, 16, 15, 16, 15 (mean: 15.4; S.D.: 0.7).

The frontal scalation of tirbaki is of the G-type and the cephalic neuromasts of the first frontal pair are of the open type, like all Aphyosemion from Gabon; no evidence of a significant sexual body dimorphism, except for fin length which is longer in male, can be disclosed, like in all Aphyosemion species (Scheel, 1968).

Because it is of utmost importance for the diagnosis of tirbaki and of coeleste (see further: systematic relationships), the specific count of the D/A deviation has been done, by the author with the microscope only, on larger samples: D/A= + 3.4 on average of 31 specimens of both sexes; standard deviation: 0.4: this value has never been found in coeleste, even the highest one, +5 in rare specimens. Moreover, in table 1, a comparison of some characteristics with their range, mean and standard deviation, is given for the above 8 types of tirbaki and for 7 specimens of coeleste collected in 1979, in the Congo, and studied then by the author. On that small samples, tirbaki and coeleste appear to be distinct by the dorsal fin ray count, by the D/A ratio, by the depth of body and peduncle, maybe also by the lateral scale count and the anal fin count.

Life colors.- Male, red vertical irregular markings on the sides on a deep blue background; the yellow margins and red submargins of the caudal are also present in the dorsal, anal and ventral fins. Female, strongly subdued as usual in the genus: overall brown with a few red irregular markings rather in lines and in the median part of sides; the caudal fin is hyaline, whereas the dorsal and anal fins are irregularly flamed with red.

Colors in preservative.- Male, with the same markings as in life, except the blue background which has disappeared after alcoholic maintenance; yellow colors are replaced by white in margins of the caudal, dorsal and anal fins; in addition, the upper sides are darkly reticulated from the opercle level up to the peduncle and the caudal fin is innerly flamed with black (i.e. lines along rays); female shows also the dark upper sides, but only until the dorsal insertion level; some red spots, arranged in discontinuous lines, can be seen on sides, irregularly depending on specimens; dorsal and anal fins with a flamed pattern along rays, basally red then black distally; the anal fin margin is thinly bordered with black.

Systematic relationships.- As stated above (the superspecies concept), A. tirbaki is diagnosed by its color pattern only, from a series of eight isomorphic vicariant species, the members of the ogoense superspecies. Among them, it presents a somewhat higher number of unpaired fin rays (D= 14-15; A= 16), a lower D/A deviation (D/A= +3 or +4) and a lower longitudinal scales count (LL= 27-28), but knowing the overlapping variability observed with these fishes, from population to population, and, from species to species (Huber, 1978), this should be confirmed on larger samples from several localities.

On the other hand, A. tirbaki is morphologically separated from the components of the coeleste superspecies which lives also in southern Gabon and Congo, sometimes sympatrically with the ogoense superspecies and both are then assumed to be non related; by color pattern, surprisingly, the tirbaki male is little separated from coeleste: the pattern of the sides is made of many red bars instead of a few red spots behind the opercle, the rest being identical (see figure 2 and 3, for comparison, and further, the discussion on color convergence).

Ecology.-The type locality is a forest stream containing crystal clear, slow- moving water and a gravel bottom. At the time of collection, the stream was 2.5 to 3.5 m wide and 30-60 cm deep, and was partially shaded by overhanging terrestrial plants and semi-aquatic plants along the edges. A. tirbaki was very abundant at the time of collection. Although it was the only Cyprinodont fish collected at this locality, it is likely that Procatopus (Plataplochilus) terveri Huber, 1981 is present in nearby areas. Other species collected were Barbus sp. (these are presently under study), Ctenopoma kingsleyae Günther, 1896, and an undetermined characoid. Other animals included freshwater shrimps and tadpoles. The water was neutral to slightly acidic and relatively soft: pH= 6.2-6.4 and hardness less than 1dGH (P. Tirbak, pers. comm.).

Based on aquarium observations, A. tirbaki lays eggs among plants located close to the surface. Incubation time is about three weeks in water temperatures of 21° C. The maximum size is 70 mm (P. Tirbak and W. Eberl, pers. comm.).

Distribution.- Aphyosemion tirbaki is thus far known only from two localities, the type locality and a second locality about 20 kms to the northwest (Blum locality BSWG 97/9 of August 1997), where the species was found sympatrically with one congener, A. lamberti. Both localities lie along the only passable road, which runs in a north-south direction. The closest known location for the superficially similar A. coeleste is the village of Mounana, about 20 km south of the type locality. It is doubtful if the range of the new species is very extensive, since not far to the north one finds a distinctively different fauna, with three different species of Aphyosemion: A. lamberti (member of the elegans superspecies), A. punctatum ( wildekampi superspecies), and A. splendidum (batesii superspecies). The comparative distribution of the ogoense and coeleste superspecies is given in figure 4 (based on the collecting localities, listed in Huber, 1996).

Companion Frontier species.-
A. tirbaki
superficially bears a close resemblance to A. coeleste. Although both species are members of the subgenus Mesoaphyosemion, they belong to different superspecies and are not intimately related.

Etymology.- The species is named in honor of one of its discoverers, Peter J. Tirbak, a renowned U.S. killi-hobbyist from Palo Alto, California.

The frontier species concept.

It has been observed in cyprinodontiform fishes that congeneric species within the same species group (i.e., superspecies) very rarely occur sympatrically (in a restricted sense; also termed syntopically, by some authors). They replace each other allopatrically, with neighboring distributions. Therefore, the distribution of a superspecies is the aggregation of the distribution of the individual members. Inversally, sympatry is common among congeners belonging to different superspecies (Huber, 1977). These instances of sympatry may involve as many as five species (Huber, 1978; 1992), structured in independant tribes scattered all over the marigot. Brosset (1982) demonstrated that during times of spawning females are responsible for mate selection among sympatric congeners, and this is based, at least in part, by recognition of subtle differences in color pattern of the unpaired fins and posterior sides of the body.

In west Africa, especially in the area from eastern Nigeria to the Congo, one notices an abrupt change in species composition between the coastal plain and the interior plateau. The superspecies are restricted to either upland or lowland areas, especially when the transition between the coastal plain (0-250m) and the plateau (350-700m) is sharp and rapid. On the narrow coastal plain, superspecies are distributed in strates parallely to the coast line, whereas on the plateau a distribution in patches is observed (Huber, unpublished). Within areas of geographical transition, upland and lowland species may occur sympatrically, although there appears to be no significant reciprocal geographic penetration. And, apart from the barrier plain-plateau, no other barrier (e.g. river drainages, soil composition, change in forest coverage, etc.) has been established to explain the limits of distribution of an individual species or of a given superspecies (Huber, unpublished): the individual sympatric species or the resulting sympatric superspecies do not show, respectively, superimposed distributions. None the less, the distributional structure of a superspecies appears to be made of two types, both in the plateau and in the plain. First, usually 1, or 2 species are largely distributed, usually near the center and with a variable, generalized color pattern (e.g. louessense in the ogoense superspecies). Second, several species are restricted in some areas of the periphery, with a much less variable and more typified color patterns (e.g. caudofasciatum, zygaima, ogoense, etc., in the ogoense superspecies). Each of these latter phenotypes is in direct conflict of expansion with a component of another allopatric superspecies. Both reject each other, mutually. No case of a species with a restricted range and a typified color pattern is known from the middle of the distribution of a given superspecies.

This was termed by me (Huber, 1980, 1981) the "frontier species" concept. Since then, lots of examples have been forwarded to strengthen the concept in Aphyosemion and also in other Cyprinodont genera and also in South America. Frontier species occurring at the periphery of the overall distribution of the superspecies possess distinctive pigmentation, which in turn may be very similar to that found, vicariantly, in a peripherally distributed species belonging to another superspecies. Males belonging to different allopatric superspecies may be superficially very similar in appearance in such cases: a color convergence results. No such a case of a color convergence is known among sympatric congeners (which belong to distinctive superspecies separated by morphological characteristics) over the overall distribution of the genus.

In the genus Aphyosemion, the following examples can be reported (figure 5). A. celiae is the frontier species of the calliurum superspecies in the Cameroun coastal plain, facing A. cinnamomeum, the frontier species of the gardneri superspecies in the plateau, both with a unique shared color pattern of the caudal fin: a crescent shape organisation of colors. A. marmoratum is the frontier species of the scheeli superspecies in the Nigeria/Cameroun coastal plain, facing A. mirabile, another frontier species of the gardneri superspecies in the plateau foothills, both with a marbled pattern of posterior sides and vertical fins. A. exigoideum is the frontier species of the striatum superspecies in the Gabon coastal plain, facing A. joergenscheeli, the component of the ogoense superspecies in the plateau, both with a circumcaudal red line on a blue-green background. A. rectogoense is the frontier species of the elegans superspecies in the Gabon/Congo savannah plateau, facing A. ogoense, the frontier species of the ogoense superspecies in the same but forested plateau, both with a basal red line on lower sides and a flamed caudal pattern.

Facing frontier species may share close similarities in color and color pattern, but display distinct differences in body size, body and fin shape, fin-ray counts, and especially D/A ratio. A. tirbaki and A. coeleste seem to provide another example of frontier species with color similarity. They differ in a number of meristic and morphological characters (Table 1), but are similarly colored, except for red vertical markings on the side of the body in tirbaki and the postopercular golden and red region in coeleste. Karyotypes of the two species have been studied by G. Collier, who states (pers. comm.), "the karyotype of A. sp. EBT96/7 is superficially similar to that of A. coeleste; however, comparison of DNA sequences of portions of mitochondrial Cytochrome b and 12S RNA genes do not support a close relation between the two species."

Much remains to be learned about the taxonomy, distribution, and ecology of the killifishes inhabiting coastal regions of tropical west Africa. Inaccessibility of much of this area precludes detailed knowledge of population dynamics in areas near the peripheries (i.e., the "frontiers") of the ranges of different species. It may be observed that the primary role of the female in mate recognition of sympatric species rules out the possibility of the sympatric occurrence of similarly colored congeners. The theoretical consequence of this is that a hypothetical past occurrence of a mutational speciation accompanied of a color convergence with a sympatric congener would have induced a strong mutual geographic rejection. With or without color convergence, the concept of frontier species in tropical Cyprinodonts is hypothesized to be linked, for these fishes bound to reclusive biotopes, to the drastic climatic and ecological variations of the early Holocene (Huber, unpublished). These ideas are conjectural, however, and at present no definitive answers are available to explain the "frontier species" phenomenon. Experimental work involving both behavior and genetic compatibility are required in order to resolve these problems, together with intensive field prospection.


Acknowledgement.- I wish to thank the following individuals for advice and information, or for reviewing an earlier draft of this paper: J. Daget (MNHN), G. Collier (University of Tulsa, Tulsa, Oklahoma), C. R. Gilbert (Florida Museum of Natural History, University of Florida, Gainesville), B. J. Turner (Virginia Polytechnic Institute and State University, Blacksburg), A. C. Radda (Institut für Virologie, Vienna). I also acknowledge and thank the collectors, T. Blum, W. Eberl, and P. Tirbak, who provided specimens for study and for sharing their field and aquarium observations. M. Chauche (Saint Ouen) and R.H. Wildekamp (Gemert) were so kind to prepare respectively the comparative map of distribution and the idealized sketches of the males of the species under study.


Brosset A. 1982. Le Peuplement de Cyprinodontes du Bassin de l'Ivindo, Gabon. Rev. Ecol. (Terre Vie), 36: 233-292, figs.

Clausen HS. 1967. Tropical Old World Cyprinodonts. Akademisk Forlag, Copenhagen, Denmark, 64 pp., 22 figs.

Hoedeman JJ. 1958. The frontal Scalation pattern in some groups of Toothcarps (Pisces, Cyprinodontiformes). Bull. Aquatic Biol., 1 (3): 23-28, 7 figs.

Hubbs CL, Lagler KF. 1958. Fishes of the Great Lakes region. Bull. Cranbrook Inst. Sci., 26: v-xi, 1-213 (revised edition).

Huber JH. 1977. Une Chaîne de deux Aphyosemion sympatriques dans les Monts de Cristal, Gabon, avec Description d'une espèce nouvelle: A. mimbon n. sp. Rev. fr. Aquariol. 4: 3-10, 13 figs., tab.

Huber JH. 1978. Contribution à la Connaissance des Cyprinodontidés de l'Afrique Occidentale: Caractères taxonomiques et Tentative de Regroupement des Espèces du Genre Aphyosemion. Rev. fr. Aquariol. 5: 1- 29, 9 figs., 30 photos, 6 maps.

Huber JH. 1980. Rapport sur la deuxième Expédition au Gabon (Août 79). Etude des Cyprinodontidés récoltés. Rev. fr. Aquariol. 7: 37-42, figs.

Huber JH. 1981. Die Überart Aphyosemion ogoense. Eine systematische Übersicht, sowie die Darstellung von hypothetischen Artbildungsmodellen. D.K.G.- Journal 13 (9): 133-148, 16 figs., 3 maps.

Huber JH. 1992. Review of Rivulus. Ecobiogeography - Relationships. Cybium Suppl. , Société Française d'Ichtyologie Publ.: 586 pp., 40 pls., 85 figs, 8 tabs, 13 maps.

Huber JH. 1996. Killi-Data 1996. Updated checklist of taxonomic names, collecting localities & bibliographic references of oviparous Cyprinodont fishes (Atherinomorpha); in french, english, and german. Cybium, Soc. fr. Ichtyologie, Ed. , Paris: 400 pp., 20 maps.

Leviton AE, Gibbs Jr. RH, Heal E, Dawson CE. 1985. Standards in herpetology and ichthyology : Part I. Standard symbolic codes for institutional resource collections in herpetology and ichthyology. Copeia 1985: 802-832.

Murphy W.J., Collier GC. 1996. Phylogenetic relationships within the Aplocheiloid fish genus Rivulus (Cyprinodontiformes, Rivulidae): Implications for Caribbean and central American Biogeography. Mol. Biol. Evol., 13 (5): 642-649, 3 figs.

Murphy W.J., Collier GC. 1997. A Molecular Phylogeny for Aplocheiloid Fishes (Atherinomorpha; Cyprinodontiformes): the Role of Vicariance and the Origins of Annualism. Mol. Biol. Evol., 14 (8): 790-799.

Myers GS.  1924. A new Poeciliid Fish from the Congo, with remarks on Funduline genera. Amer. Mus. Novitates, 116: 9pp., 4 figs.

Parenti LR. 1981. A Phylogenetic Analysis of Cyprinodontiformes Fishes (Teleostei; Atherinomorpha). Bull. Am. Mus. Nat. Hist., 168: 335-557.

Pellegrin J. 1930. Poissons de l'Ogooué, du Kouilou, de l'Alima et de la Sangha recueillis par M. A. Baudon. Description de cinq Espèces et cinq Variétés nouvelles. Bull. Soc. Zool. France 55: 196-210, fig.

Radda AC. 1977. Vorläufige Beschreibung von vier neuen Subgenera von der Gattung Aphyosemion Myers. Aquaria, 24: 209-216, 1 fig.

Romand R. 1992. Cyprinodontidae. In: Lévêque C, Paugy D, Teugels GG. Faune des Poissons d'eaux douces et saumâtres d'Afrique de l'Ouest. Arial; Tome 2. Orstom/Mrac Ed. Faune tropicale XXVIII: 586-654.

Wildekamp RH, Romand R, Scheel JJ. 1986. Cyprinodontidae. In: Daget J., Gosse JP , Thys van den Audenaerde D, eds. Checklist of the freshwater Fishes of Africa (C.L.O.F.F.A. 2). Orstom, Paris: 165-276.

Scheel JJ. 1968. Rivulins of the Old World. T.F.H. publ., New Jersey, 480pp.

Zee J.R. van, Wildekamp RH. 1995. Description of a new Fundulopanchax species (Cyprinodontiformes; Aplocheilidae) from the Niger Delta, with a Redescription of the genus Fundulopanchax. J. Afr. Zool., 108 (5), 1994: 417-434, 8 figs.


Laboratoire d'Ichtyologie générale, Museum national d'Histoire naturelle, 43, rue Cuvier, 75231 Paris Cedex 05, France


Table 1. Comparison of selected morphometric and meristic characters between specimens of Aphyosemion tirbaki and Aphyosemion coeleste (range, then between parenthesis, mean and standard deviation; measurements, expressed in hundredths of standard length).

                                                                        A. tirbaki                                               A. coeleste 

No. of specimens                                                    8                                                      7

Standard length (mm)                                              29.5-38.5                                          20.3-30.4

Total length (%)                                                      121-125                                            117-125

                                                                             (123.25; 1.35)                                   (121.50; 2.57)

Predorsal length (%)                                               62-67                                               62-67

                                                                             (64.50; 1.58)                                     (64.29; 1.91)

Body depth at anal origin (%)                                   20-22                                               18-20

                                                                             (21.10; 0.62)                                     (19.36; 0.58)

Depth at peduncle level (%)                                     12-13                                               10-11

                                                                             (12.50; 0.48)                                     (10.57; 0.47)

Head length (%)                                                      24-30                                               23-27

                                                                             (27.14; 1.73)                                     (24.80; 1.38)

Dorsal-fin rays                                                        14-15                                               10-12

                                                                             (14.50; 0.50)                                     (10.9; 0.6)

Anal-fin rays                                                           16-17                                               14-15

                                                                             (16.25; 0.43)                                     (14.6; 0.5)

D/A deviation ratio                                                   +3 to +4                                           +7 to +9

                                                                             (3.5; 0.5)                                          (8.1; 0.6)

Lateral line scales                                                   26-28                                               29-31

                                                                             (27.4; 0.7)                                         (30.1; 0.6)

Transversal body scales                                          8-9                                                   7-8

                                                                             (8.1; 0.3)                                          (7.7; 0.5)


Figure 1: idealized sketches of males of the ogoense superspecies. R.H Wildekamp.

A. tirbaki from Tsotandzala, southeastern Gabon.

A. ogoense from Banbama, southern Congo.

A. ottogartneri from Lutété, southern Congo.

A. pyrophore from Mpoukou, southern Congo.

A. caudofasciatum from Ogowé, southern Congo.

A. louessense from Lékoli Magogo, southwestern Congo.

A. zygaima from Mindouli, southern Congo.

A. labarrei from the Madimba area, southern Zaïre.

A. joergenscheeli from Mbigou, southwestern Gabon.

Figure 2: adult male, A. tirbaki topotype, not preserved. R. Lütje.

Figure 3: adult male, A. coeleste from Malinga, southern Gabon (for comparison). M. Chauche.

Figure 4: map of collecting sites of the Aphyosemion ogoense superspecies and of the coeleste superspecies. M. Chauche.

Figure 5: idealized sketches of males of "pairs" of frontier species with color convergence. R.H. Wildekamp.

A. tirbaki from Tsotandzala, southeastern Gabon.

A. coeleste from Massango, southern Gabon.

A. celiae from Badouma, western Cameroun.

A. cinnamomeum from Nguti, western Cameroun.

A. marmoratum from West of Mbongé, western Cameroun.

A. mirabile from Mbio, western Cameroun.

A. exigoideum from Mandilou, western Gabon.

A. joergenscheeli from Mbigou, southwestern Gabon.

A. rectogoense from 31km East of Franceville, southeastern Gabon.

A. ogoense from Banbama, southern Congo.

Additional reading of this author's publications ? Select any ONLINE article.