Phylogenetic position and tentative generic placement for Cyprinodon martae Steindachner , 1875 ( Teleostei : Cyprinodontiformes ) , a killifish from northern Colombia

Examination of photographs and a radiograph of the holotype of Cyprinodon martae, a rare killifish from coastal northern Colombia, revealed that it is not a rivulid by not exhibiting any diagnostic apomorphic condition of the Rivulidae, as well as the character states diagnosing the suborder Aplocheiloidei are not present. On the other hand, besides possessing many characters common among cyprinodontoids but not found among aplocheiloids, it has diagnostic features of the suborder Cyprinodontoidei, such as neural and hemal spines of the preural vertebrae 3 distinctively wider than those spines of the anterior vertebrae. These characters associated to the jaw dentition morphology unambiguously support its inclusion among the Cyprinodontidae, whereas the presence of an enlarged supraoccipital crest strongly suggests that is a member of the Caribbean killifish genus Cubanichthys.


Introduction
Cyprinodon martae Steindachner, 1875 was first de scrib ed on the basis of material collected in Santa Marta, northern coast of Colombia (Steindachner, 1875), the only type specimen being deposited in the Na tur his to ri sches Museum, Wien (NMW 76519).Curiously, this pe cu liar killifish species, with long dorsal and anal fins never found among species of Cyprinodon Lacepède, 1803 (Cy pri no dontidae, Cyprinodontoidei) as today delimited, was not collected again.However, some controversy about the possible correct group placement of this species has sporadically appeared in the literature, besides unpublished rumours that it is not even a cyprinodon tiform fish (WiLdekamp, 1995).
Garman (1895) kept C. martae in Cyprinodon, but warning for the shape of the body most similar to Le bias (= Aphanius nardo, 1897, Cyprinodontidae, Cy pri no dontoidei) or Jenynsia Günther, 1866 (Anablepidae, Cy prinodontoidei).FoWLer (1939), considered C. martae possibly related to Cyprinodon jamaicensis FoWLer, 1939 (= Cyprinodon riverendi (poey, 1860) without justification, whereas WiLdekamp (1995) listed Cyprinodon mar tae as a nomen dubium in the synonymy of Cyprinodon dearborni meek, 1909 from coastal Venezuela and adjacent Caribbean islands, based only on geographical proximity, but not considering the different characters exhibited by the two species, including among others the short dorsal and anal fins of C. dearborni. huber (2000) after examining a photograph and a radiograph of the holotype, showed that it could not be assigned to Cyprinodon.Subsequently and more unexpectedly, huber (2004) considered it as possibly being an annual rivulid (Rivulidae, Aplocheiloidei) of a new distinctive genus.More recently, it has been reported in aquarium and ichthyological websites that huber (2012, not seen) described a new genus for C. martae in his private journal, placing it close to Terranatos taphorn & thomerSon, 1995 (Rivulidae, Aplocheiloidei) (e.g., http://www.killidata.org/zzmartaeCyprin.php).Since characters provided in the original description of C. mar tae highly confl icting with its placement among rivu lids according to our present knowledge about morpho logy of cyprinodontiforms, photographs in different views and a radiograph image of the type were requested for study, resulting in the present paper.

Material and methods
The specimen studied is deposited in Naturhistorisches Museum, Wien (NMW 76519).Comparative material is listed in coSta (2012).Osteological nomenclature fol lows coSta (2006).

Results and Discussion
The examination of photographs and a radiograph image of the holotype (Fig. 1) associated to characters described in the original description indicate that C. martae: -1) is a cyprinodontiform, by having a single bladelike epural and a welldeveloped neural spine of the preural vertebra 2, making the dorsal and ventral portions of the caudal skeleton strongly symmetrical (Figs.1B, F), which together comprise a morphological complex of derived character states that unambiguously diagnoses the Cyprinodontiformes (roSen, 1964;parenti, 1981;coSta, 2012); 2) is not a rivulid, by not having a tubular anterior naris (Fig. 1B), which is a synapomorphy of the Aplocheiloidei (parenti, 1981;coSta, 1998), a welldeveloped preopercular channel (Fig. 1B), instead of a rudimentary channel, restricted to a short shallow segment on the dorsal portion of preopercle, which is a synapomorphy of the Rivulidae (parenti, 1981;coSta, 1998), the branchiostegal membrane free of the isthmus (Fig. 1G), instead of continuous, which is a synapomorphy of the Rivulidae (parenti, 1981;coSta, 1998), besides having an intense silver colouration on the opercle and trunk (Figs.1A-B) and a distinctively long anterior proximal radial of the anal fin (Fig. 1E), which are conditions common among cyprinodontoids and other atherinomorphs, but not occurring in aplocheiloids; -3) is a cyprinodontoid, by having the neural and hemal spines of the preural vertebrae 3 distinctively wider than those spines of the anterior vertebrae (Fig. 1F), which is a synapomorphy of the Cyprinodontoidei (coSta, 2012: character state 9.1), hypurals completely fused, forming a single broad plate, with dorsal and ventral margins forming an angle of about 90 o (vs.about 70 o or less), a condition uniquely found among the Cyprinodontoidei; -4) is a cyprinodontid, by having tricupidate teeth on jaws (Steindachner, 1875), which is synapomorphic both for the Anablepidae (parenti, 1981;coSta, 1998;Ghedotti, 1998) and to the Cyprinodontidae (parenti, 1981;coSta, 2011), and by possessing a single row of teeth on jaws, since the Cyprinodontidae differs from all other cyprinodontiforms by having a single tooth row, often followed by an internal irregular row of a few, smaller teeth (coSta, 2011); -5) probably is a member of the genus Cubanichthys hubbS, 1926 (Cyprinodontidae, Cyprinodontoidei), by having an enlarged supraoccipital crest (Fig. 1D), which is a synapomorphy of Cubanichthys, the only genus of the Cubanichthyiinae (parenti, 1981).In addition, the relative position of dorsal and anal fins, with the dorsal fin origin placed anteriorly to the analfin origin, which is posterior to the midlength between snout and caudal fin base (Figs.1A, C); vs. anterior when dorsalfin origin is anterior to the analfin origin) is a condition common among cyprinodontids, but never found in rivulids; the presence of a rounded snout with a subdorsal mouth opening (Fig. 1B, D); vs. dorsal mouth opening is a condition found in Cubanichthys, but never in rivulids.
Since frontal scales are missing in the holotype (Fig. 1H), it is not possible to check the presence of a large central Escale, which is a synapomorphy of Cubanichthys (coSta, 1997).
Cubanichthys, a generic name not available at the time of the original description of C. martae, presently comprises two species from the Caribbean islands, C. cuben sis (Eigenmann, 1903) from Cuba and C. pengelleyi (Fowler, 1939) from Jamaica.Thus it would not be sur-prising C. martae from the Caribbean area of Colombia being a member of Cubanichthys, but its definitive generic allocation should be conducted only after confirmation is made through the study of additional specimens.