Pterolebias Garman

Wilson J. E. M. Costa, 2005, The Neotropical annual killifish genus Pterolebias Garman (Teleostei: Cyprinodontiformes: Rivulidae): phylogenetic relationships, descriptive morphology, and taxonomic revision., Zootaxa 1067, pp. 1-36 : 20-24

[[ Genus Pterolebias Garman ]]

Discussion

Taxonomy. Pterolebias possibly poses the greatest taxonomic problems within the Rivulidae (see History above). This is due to the wide geographic range oîP P. longipinnis , which is unique for an annual fish, and overlaps the type locality of two similar, controversial species, P. bokermanni and P. luelingi .

Travassos (1955) distinguished P. bokermanni from P. longipinnis by the former having pelvic-fin bases medially united, dorsal and anal fins placed more posteriorly, a slender body, pelvic fin longer, fewer scales in the transverse series, more scales in the longitudinal series, and greater development of cephalic “pores” (i.e. neuromasts). However, Travassos never examined any specimens of P. longipinnis from the type locality or from any other region, and based comparisons on the limited data provided in Garmans (1895) description. None of these features differs among populations formerly identified as P. longipinnis , P. bokermanni , and P. luelingi .

The reasons that P. luelingi was first placed in Rivulichthys by Meinken (1969) and not in Pterolebias are unknown. The few characters presented in the original description of R. luelingi are congruent with those included in the original description of P. longipinnis . Thomerson (1984) considered R. luelingi to be a synonym of P. longipinnis , although he recorded eight pelvic-fin rays for the types of P. longipinnis , in contrast to seven rays recorded by Meinken for R. luelingi . However, Schindler (2004) used this putative difference in number of pelvic-fin rays to distinguish P. luelingi , occurring in the Mamoré River basin, from P. bokermanni , from the Guaporé and Paraguay river basins, and from P. longipinnis . In fact, the type locality of P. bokermanni is Guajará-Mirim, on floodplains of the Mamoré River, although it is also recorded from the Guaporé River floodplains (Travassos, 1955). All of the approximately 400 specimens of Pterolebias examined in the present study have eight pelvic-fin rays, except for a few specimens with nine rays. Schindler (2004) also noted a reduced squamation on the anal-fin base in males, including two or three rows of scales in P. longipinnis and P. luelingi , in contrast to four or more rows in P. bokermanni . However, all adult males of Pterolebias examined during the present study have four to seven rows, except P. phasianus , which has a single irregular row. Therefore, diagnostic features recorded by Schindler (2004) are not useful for distinguishing P. luelingi and P. bokermanni from P. longipinnis .

Characters examined were not useful for recognizing P. bokermanni and P. luelingi as valid species, nor for distinguishing geographically distant populations previously identified as P. longipinnis from the Paraná -Paraguay system. Some distinctive color patterns, such as stripes on male anal fin, were noted in individuals from the Mamoré and Paraguay river basins in Brazil (see above description). However, these color features are not constant, are absent from a large percentage of individuals from those regions, and therefore are not useful as diagnostic characters in this particular instance.

Phylogenetic relationships. Both monophyly of Pterolebias and the sister group relationship between Pterolebias and Gnatholebias are strongly supported by bootstrap values (Fig. 1). The clade Pterolebias was previously corroborated by morphological data (Costa, 1998a) and molecular studies (Murphy et al., 1999; Hrbek & Larson, 1999). However, the clade Pterolebias plus Gnatholebias , already reported in morphological analysis (Costa, 1998a), is in conflict with molecular phylogenetic studies, which strongly support the monophyly of a clade including Gnatholebias and other taxa endemic to northern South America (Murphy et al., 1999; Hrbek & Larson, 1999) (Fig. 11).

Monophyly of Pterolebias is herein supported by four unambiguous synapomorphies: ventral process of angulo-articular narrow and pointed (16.2) (Figs. 3a, c), medial flap of second pharyngobranchial expanded (33.1) (Fig. 3g), small metallic orange spots on humeral region in males (82.1) (Figs. 5, 6, 9), and bars on pectoral fin in males (90.2) (more conspicuous in the example of Fig. 9). The Pterolebias clade is further corroborated by four synapomorphies homoplastically occurring in other rivulids: maxilla greatly twisted (13.2) (Figs. 3a-b), dorsal portion of metapterygoid strongly constricted (19.2) (Fig. 3a), posterior process of quadrate long (20.1) (Fig. 3a), and basihyal long (25.1) (Fig. 3a); and one reversal: pectoral fin rounded (62.0).

Monophyly of the clade comprising Pterolebias and Gnatholebias is supported by three unambiguous synapomorphies: interarcual cartilage minute (31.1) (Figs. 3d, g), anterior proximal radiais of anal fin curved and posteriorly directed (48.1) (Fig. 4b), and anal-fin base with scales (75.1). This assemblage is also supported by four synapomorphies homoplastically occurring in other rivulid lineages: external median teeth of premaxilla and dentary displaced laterally (17.1) (Figs. 3a-c), ischial process of pelvic girdle vestigial (55.1) (Fig. 4c), eight pelvic-fin rays (56.1), and pelvic fin long in males (64.1) (Figs. 5, 6, 9, 12). However, in a study of mitochondrial DNA by Hrbek & Larson (1999), Gnatholebias zonatus was hypothesized to be more closely related to Rachovia stellifer than to other rivulids, and Terranatos dolichopterus was hypothesized to be the sister group of the clade including G. zonatus plus R. stellifer (Fig. 11b). A similar hypothesis was obtained from another mitochondrial DNA study (Murphy et al., 1999), in which G. zonatus was considered to be the sister group of T. dolichopterus , but R. stellifer was not included in the analysis (Fig. 11a).

I have found no morphological support for an assemblage including Gnatholebias zonatus , Terratanos dolichopterus , and Rachovia stellifer , since all apomorphic conditions shared by these three genera are present in several other rivulid lineages. Both R. stellifer and T. dolichopterus have the pelvic fin slightly elongated in males, which could be considered homologous to the long pelvic fin of G. zonatus and G. hoignei . However, the pelvic fin is much longer in Gnatholebias and Pterolebias . In addition, in the latter two genera, the pelvic fin elongation is due to a long fourth ray (of eight rays) always occurring in the fin. In T. dolichopterus , however, the longer ray is the third (of eight), thus not considered a homologous condition; and in R. stellifer the longest ray is the fourth, but since there are only seven pelvic-fin rays in that species homology is doubtful. Gnatholebias zonatus and T. dolichopterus share a medial fusion on the pelvic fin (65.2), but this is not significant when compared to the many morphological synapomorphies shared by Gnatholebias and Pterolebias (see above). Gnatholebias zonatus and T. dolichopterus share reduction or loss of the ischial pelvic process (55.1), but this condition is also present in Pterolebias . Like Gnatholebias species, T. dolichopterus has filamentous rays on the unpaired fins (67.1, 68.1), but this feature is also present in Pterolebias and in some species of Rachovia and Austrofundulus (e. g., R. brevis (Regan) , A. leohoignei Hrbek, Taphorn & Thomerson , A. leoni Hrbek, Taphorn & Thomerson ), which would be more closely related to Terranatos according to the morphological studies. In addition, T. dolichopterus , R. maculipinnis , and A. transilis share four synapomorphies: dorsally directed process on the base of anterior epipleural ribs (40.1), dorsal profile of the anterior portion of body convex (59.1), supraorbital spot adjacent to eye (85.1), and dark pigmentation reduced on flank and fins in females (92.1). The first feature, already described and illustrated by Costa (1998a: fig. 28a), seems to be unique among aplocheiloid fishes.

Although morphological and molecular studies often produce slightly different estimates of phylogenetic relationships among species of the same group, Hillis & Wiens (2000) considered significant incongruence rare, pointing out some common causes, such as undersampling of characters or taxa, differences in phylogenetic methods and differences in rooting techniques. However, none of these causes seems to be applicable, since all studies are based on parsimony analysis, similar outgroup analysis methods, and considerable numbers of characters and representative taxa, with conflicting nodes having high bootstrap values.

It still not possible to satisfactorily understand the incongruence found between morphological and molecular hypotheses of relationships among Pterolebias , Gnatholebias and other rivulids, making it difficult to establish which data set is misleading. It would be desirable to improve estimates of phylogenetic relationships through the addition of characters from unsampled structures in the morphological analysis or introduction of new genes in the molecular analysis, but this is beyond the scope of the present study.

Classification. Based upon a morphological phylogenetic analysis of the Rivulidae , Costa (1998a) proposed a new classification, in which Pterolebias was divided into four genera ( Pterolebias , Gnatholebias , Aphyolebias , and Micromoema ). The groups recognized as genera were corroborated by subsequent molecular phylogenetic studies (Murphy et al, 1999; Hrbek & Larson, 1999), and the general classification has been employed both in scientific and aquaristic literature. However, Schindler (2004) considered Gnatholebias to be a junior synonym of Pterolebias , claiming that Pterolebias plus Gnatholebias form a monophyletic group, and that “... it would be better to adopt a more conservative nomenclature than to change the systematic categories after every new analysis ...” (Schindler, 2004: 74). This proposal is not followed here by the following reasons:

1) Pterolebias including species placed in Pterolebias plus Gnatholebias sensu Costa (1998a) is not a conservative nomenclature as proposed by Schindler (2004), since this generic composition would be new. Pterolebias was a more inclusive genus prior to publication of Costas (1998a) work, and included species today placed in Aphyolebias , Moema , and Micromoema , thus forming a paraphyletic assemblage, as estimated both by morphological and molecular studies (Costa, 1998a; Murphy et al., 1999; Hrbek & Larson, 1999). Therefore, Pterolebias including species placed in Gnatholebias but excluding species of Aphyolebias , Moema , and Micromoema should not be viewed as conservative, especially considering that Costas classification has been consistently employed during the last seven years.

2) Although the clade Pterolebias plus Gnatholebias is strongly supported by morphological characters (Costa, 1998a; the present study), it is still controversial, since it is not corroborated in two independent molecular phylogenetic analyses (Murphy et al., 1999; Hrbek & Larson, 1999) (Fig. 11). Thus, Pterolebias , as proposed by Schindler (2004), would be paraphyletic according to molecular studies.

3) Both Pterolebias and Gnatholebias sensu Costa (1998a) are monophyletic assemblages, well corroborated by morphological and molecular studies (Costa, 1998a; the present study). In addition, Pterolebias and Gnatholebias are readily distinguished even by external morphological characters, easily accessible both by museum systematists and amateur ichthyologists (e. g., short snout, rounded pectoral fin, and dark bars on the pectoral fin in Pterolebias , vs. long snout, pointed pectoral fin, and pectoral fin with black ventral margin in Gnatholebias ) (Fig. 12).

4) These two monophyletic groups (i. e., Pterolebias and Gnatholebias ) are potential indicators of historic biogeography, since, although not necessarily sister groups, they are closely related taxa with completely disjunct biogeographic distribution patterns.