Systematics of the genus Cnesterodon Garman, 1895 (Cyprinodontiformes: Poeciliidae: Poeciliinae)

, a new speciesfrom the rio Ribeira de Iguape, described herein, and a new species from the Rio Grande do Sul, being described by in a separat epaper. Diagnoses and distribution ranges are provided for each species as well as a key to identification of the species.Uma hipotese de relacoes filogeneticas e proposta para as especies de


Introduction
The genus Cnesterodon Garman, 1895 comprises small poeciliids distributed in southern South America. Cnesterodon species live in the upper rio Araguaia basin, the Paraná-Paraguay system, the rio Uruguay basin, and along coastal drainages from São Paulo to Argentina, as well as in small drainage basins of western Argentina (Fig. 1).

Materials and Methods
The following counts were taken under a stereomicroscope: (a) predorsal scales: number of scales from posterior margin of large scale on dorsum of head to first dorsal-fin ray, excluding modified scale that may be present at origin of dorsal fin; (b) longitudinal series scales: counted from the second scale above pectoral fin insertion, to the end of hypural plate; (c) circumpeduncular scales; (d) scales in transverse row between origin of dorsal and anal fins, including minute scales present in some specimens at the base of anal fin; (e) number of epipleural ribs; (f) number of pleural ribs; (g) number of vertebrae: the urostyle is counted as one element; (h) number of fin rays: the last two rays in dorsal and anal fins were counted as independent elements. Gonopodial rays were counted in cleared and stained specimens. Rudimentary and procurrent rays were included in pectoral and caudal fin counts, respectively. All counts were made on the left side of adult specimens whenever possible. Adult males were considered those with a fully developed gonopodium. In species descriptions, numbers in square brackets following the counts indicate number of specimens for each count. Primary types counts are indicated by an asterisk. Morphometric data were obtained as distances between homologous landmarks on the lateral left surface of head and body. Specimens were digitalized as image files. Cartesian landmark coordinates were obtained using the software "TpsDig" version 1.37 by F. J. Rohlf (2003). Landmark distances were calculated using the software "LMDis" version 1.0 by R. E. Reis (1996). The table of descriptive morphometrics were elaborated with "Datax" version 4.2 by R. E. Reis and N. F. Fontoura (1993).
Thirteen landmarks were digitalized per specimen: (1) premaxillary tip, (2) supraoccipital tip, (3) origin of dorsal fin, (4) end of dorsal fin base, (5) base of middle caudal-fin rays, (6) point in dorsal profile of caudal peduncle, (7) point in ventral profile of caudal peduncle, (8) origin of anal fin, (9) end of anal-fin base, (10) base of first pelvic fin ray, (11) anterior margin of orbit, (12) posterior margin of orbit, (13) posterodorsal corner of opercle. Landmarks 6 and 7 were solely taken in order to measure caudal peduncle depth in the point of smallest depth. From these landmarks, the following measures were calculated: (a) standard length (SL), (b) head length (HL), (c) snout-occipital distance, (d) predorsal distance, (e) snout length, (f) orbital diameter, (g) dorsal-fin base length, (h) postorbital length, (i) anal-fin base length, (j) body depth, (l) pre-pelvic length, (m) pre-anal length, (n) post-anal length, (o) caudal peduncle depth. Measures, other than SL, are expressed as percents of SL, except those that are sub-units of the head, which are expressed as percents of HL. Clearing and staining followed the method of Taylor & Van Dyke (1985). In species descriptions, numbers in square brackets following the counts indicate number of specimens for each count. The entries under examined material for each species follow the sequence: Country, "Estado", "Provincia" or "Departamento", institutional abbreviation, catalogue number, total number of examined specimens in the lot (number of cleared and stained specimens is indicated by an asterisk and separated from total number by a bar), type status, county, Type-localities are listed as in original descriptions. Description of all current valid species are not provided since they have been recently described or re-described in details. Anatomical illustrations were prepared from sketches of structures from cleared and stained specimens as viewed through a camera lucida mounted on a dissecting stereomicroscope. Number and disposition of cephalic pores follow the nomenclature of Rosen & Mendelson (1960), Gosline (1949) and Parenti (1981). Only adult individuals have been examined to avoid description of ontogenetic variation. Nomenclature of the gonopodium follows Rosen & Gordon (1953). Descriptions of gonopodial morphology are based on fully developed gonopodia of large adult males. Osteological nomenclature adopted follows Rosen & Bailey (1963) and Parenti (1981). Character state assignments, transformation series and clade numbers follow the phylogenetic analysis performed by Lucinda & Reis (2005). All transformation series were considered unordered. Maximum parsimony analyses were undertaken using the mh*; bb* algorithm of Hennig86. Character optimization followed accelerated transformation model (ACCTRAN) Institutional abbreviations follow Leviton et al. (1985), except for MNHCI: Museu de História Natural do Capão da Imbuia, Curitiba and MZUEL: Museu de Zoologia da Universidade Estadual de Londrina, Londrina.

Synapomorphy List and Phylogenetic Reconstruction.
A phylogenetic hypothesis of relationships among Cnesterodon species is presented in the cladogram of Fig. 2. This hypothesis is part of a more inclusive phylogenetic study on the relationships among poeciliine genera, and the phylogenetic position of Cnesterodon in the subfamily Poeciliinae has already been discussed by Lucinda & Reis (2005: fig. 1). Synapomorphies and autapomorphies are presented and discussed below. Uniquely derived and unreversed features are indicated by two asterisks (e.g. 37-2**).

Genus Cnesterodon Garman, 1895
Clade [ 14) (Fig. 1). (1895)  Diagnosis. Cnesterodon iguape is diagnosed by the following autapomorphy: a large post-gonopodium blotch on the ventral profile in adult males. Furthermore, C. iguape is distinguished from its congeners by distal portion of anal-fin proximal radials 2 and 3 in adult males fused . Cnesterodon iguape is readily distinguished from C. omorgmatos and C. raddai by dark brown blotches along body sides (forming bars vs. circular or irregu- hypselurus. Cnesterodon iguape is distinguished from C. carnegiei, C. septentrionalis, and C. hypselurus by the presence of dark bars on sides of body vertically short, mostly confined to midline, covering less than three scales in a transverse row, never extending to dorsal and ventral profiles. Cnesterodon iguape is distinguished from C. decemmaculatus by the configuration of the bony style at gonopodium tip in adult males (relatively long and slightly arched, and bearing a membrane progressively narrowing towards tip forming a distal filament vs. relatively short and very arched. Membrane on bony style never forming a distal filament in C. decemmaculatus) (Fig. 6). Table 2. Range of SL: 18.5 to 24.9 mm (females); 18.5 to 22.4 mm (males). Body compressed, width in predorsal region uniform, about half body depth. Post-dorsal region compressed towards caudal peduncle. Predorsal profile convex. Dorsal-fin base convex. Postdorsal profile slightly concave. Pre-anal profile very convex. Anal-fin base oblique. Post-anal profile concave in females, and almost straight in males. Dorsal fin with semicircular border, located posteriorly to mid-body. Origin of dorsal fin in females on vertical passing through base of third analfin ray; in males, origin of dorsal fin posterior to vertical passing through origin of anal fin. Pectoral fin with high insertion in horizontal passing through orbital center. Longest ray reaching about seventh scale in longitudinal series. Pelvic fin small, pointed and not reaching origin of gonopodium in adult males; not reaching origin of anal fin in females. Anal fin of females with straight border. Origin of anal fin of females closer to caudal peduncle than to snout tip. Origin of anal fin of males closer to snout tip than to caudal peduncle. Gonopodial complex composed of 10 gonactinosts. Functional gonapophyses absent. Gonactinosts 2, 3, 4 fused. Gonactinost 4 with wing-like expansions. Ligastyle absent. Gonopodium symmetrical. Eight gonopodial rays. Rays 1 and 2 unbranched and short. Ray 1 with 5-7 segments. Ray 2 with 7 or 8 segments. Ray 3 with 23 or 24 segments. At tip, long slender bony style bearing narrow membrane produced in terminal filament. Four or five paired retrorse spines on distal segments of ray 4p. Ray 5a with subdistal and discrete dorsal curvature and terminated by retrorse claw. Dorsal convexity located between segments 11 to 19 of ray 5p. Rays 6, 7 and 8 branched. Distal segments of rays 6 and 7 partially ankylosed.

Color in alcohol.
Eye black with greenish brown pupil. Background color cream yellow. Scale borders and subjacent skin replete with brown chromatophores, conferring reticulate pattern to body sides. Dorsum darker than ventral region. Head dorsum dark brown. Median dark brown line along predorsal surface. Median dark brown line along preanal surface (inconspicuous in large females). Fins hyaline. Fin rays with two rows of brown chromatophores on each side, along entire ray. Seven to nine dark brown vertical bars along body sides, mostly confined to midline. Large blotch on each side of ventral portion of body near gonopodium, meeting midventral post-anal line.
Distribution. Known only from the type-locality in the upper rio Iporanga (Fig. 1).
Etymology. Cnesterodon iguape is named after the rio Ribeira de Iguape, in whose headwaters is the type-locality.

C. hypselurus + C. septentrionalis + C. brevirostratus + C. carnegiei + C. omorgmatos Clade [98]
Diagnosis. Members of this clade share the following not uniquely derived and/or reversed features: (1) first proximal radial of dorsal fin located between neural arches of 12 th and 13 th vertebrae in adult females ; and (2)  (2) bony style of ray 3 of gonopodium of fully developed mature males curly-bracket shaped (Lucinda & Garavello, 2001; fig. 6). Furthermore, C. hypselurus is readily distinguished from C. omorgmatos and C. raddai by dark brown blotches along body sides (forming bars vs. circular or irregular, respectively). Cnesterodon hypselurus is distinguished from C. decemmaculatus, C. brevirostratus, and C. iguape by dark bars of body very elongate reaching dorsal and ventral profiles, covering more than four scales in a transverse row (vs. dark bars on sides of body, mostly confined to midline, covering less than three scales in a transverse row, never extending to dorsal and ventral profiles). The absence of small scales covering lateral and ventral region of body below pectoral fin and the pointed snout also differentiate C. hypselurus from C. brevirostratus. Cnesterodon hypselurus is distinguished from C. septentrionalis, C. omorgmatos, and C. decemmaculatus, by the number of epipleural ribs (10-12 vs. 6 in C. septentrionalis, 7 or 8 in C. omorgmatos and C. decemmaculatus).
Diagnosis. Cnesterodon septentrionalis is diagnosed by the following autapomorphies: (1) six epipleural ribs; and (2) bony style of ray 3 of gonopodium of fully developed mature males large and very curved upwards, almost describing a semicircular outline, and invested with few membranous tissue (Rosa & Costa, 1993;fig. 13). Cnesterodon septentrionalis is readily distinguished from its congeners by the branched second pelvic-fin ray in adult males [42-0]; pleural ribs not associated with haemal arches in males [60-0]; first proximal radial of dorsal fin located between neural spines of 11 th and 12 th vertebrae in adult males ; first proximal radial of dorsal fin in adult females located between neural arches of vertebrae 11 th and 12 th [63-3]; less than nine caudal-fin rays entirely lying over on hypural plate [132-0]. Furthermore, Cnesterodon septentrionalis is distinguished from C. omorgmatos and C. raddai by dark brown blotches along body sides (forming bars vs. circular or irregular, respectively). Cnesterodon septentrionalis is distinguished from C. decemmaculatus, C. brevirostratus, and C. iguape by dark bars of body very elongate reaching dorsal  and ventral profiles, covering more than four scales in a transverse row (vs. dark bars on sides of body, mostly confined to midline, covering less than three scales in a transverse row, never extending to dorsal and ventral profiles). The absence of small scales covering lateral and ventral region below pectoral fin and the pointed snout also differentiates C. septentrionalis from C. brevirostratus. The absence of a longitudinal dark brown band along flank differentiates C. septentrionalis from C. hypselurus. Cnesterodon septentrionalis differs from C. carnegiei by the number of caudal fin rays (20-22 vs. 25-26, respectively), by the number of longitudinal series scales (25-26 vs. 28-31, respectively) and by the number of transverse rows between dorsal and anal fin (7 vs. 8-10, respectively).

Cnesterodon brevirostratus
Diagnosis. Cnesterodon brevirostratus is diagnosed by the following autapomorphies: (1) snout blunt; (2) presence of small scales covering lateral and ventral region of body below pectoral fin in adult females; (3) bony style of ray 3 of gonopodium of fully developed mature males large and sinuous, covered with few membranous tissue (Rosa & Costa, 1993;fig. 11). Furthermore, C. brevirostratus is readily distinguished from C. omorgmatos and C. raddai by dark brown blotches along body sides (forming bars vs. circular or irregular, respectively). Cnesterodon brevirostratus is distinguished from C. carnegiei, C. hypselurus, and C. septentrionalis by dark bars on sides of body, mostly confined to midline, covering less than three scales in a transverse row, never extend-ing to dorsal and ventral profiles (vs. dark bars of body very elongate reaching dorsal and ventral profiles, covering more than four scales in a transverse row). The absence of a longitudinal dark brown band along flank differentiates C. brevirostratus from C. hypselurus.
Distribution. Cnesterodon brevirostratus inhabits the upper portion of the rio Pelotas and rio Canoas of the rio Uruguay drainage, the rio Jacuí, the headwaters of the rio Maquiné in the Tramandaí system, and the headwaters of rio Itajaí-Açu drainage.
Remarks. The color pattern of C. brevirostratus is quite variable. It can be a typical Cnesterodon pattern with vertical bars along flanks, a pattern of bars irregularly set and shaped, or even dotted.

C. carnegiei + C. omorgmatos Clade [86]
Diagnosis. Members of this clade share the following not uniquely derived and/or reversed feature: medial surface of ascending process of premaxilla angled laterally at proximal end, forming a triangular space between proximal ends of ascending processes . Distribution. Cnesterodon omorgmatos is known solely from the type-locality in the rio Iguaçu basin (Fig. 1).