Redescription of Astyanax obscurus (Hensel, 1870) and A . laticeps (Cope, 1894) (Teleostei: Characidae): two valid freshwater species originally described from rivers of Southern Brazil

Astyanax obscurus and A . laticeps are redescribed. Both species were originally described from the laguna dos Patos system, Rio Grande do Sul, Brazil. The deepest and most robust body area close to the middle of the pectoral fins length, robust head, snout short and abrupt, body depth smaller than 41% of SL, reduced number of branched anal-fin rays, presence of one or two humeral spots, and a dark midlateral body stripe extending to the tip of the middle caudal-fin rays include both in the A . scabripinnis species complex. The presence of two humeral spots ( vs . one), external row of the premaxillary with pentacuspid teeth ( vs . tricuspid), and absence of bony hooks in all fins in males ( vs . presence in anal and ventral fins) distinguish A . obscurus from A . laticeps . These species distinguish from species of this complex by meristic and morphometric characters and color pattern. An identification key for the species of genus occurring in the laguna dos Patos system is presented. Astyanax obscurus e A . laticeps são redescritas. Ambas espécies foram descritas do sistema da laguna dos Patos, Rio Grande do Sul, Brasil. A maior altura e robustez do corpo na vertical que passa pela metade do comprimento das nadadeiras peitorais, cabeça robusta, focinho curto abrupto, altura corpo que 41% ramificados até a extremidade dos raios da nadadeira caudal permitem sua inclusão no complexo de espécies A . scabripinnis . A presença de duas manchas na região umeral ( vs. uma única), fileira externa do pré-maxilar com dentes pentacuspidados ( vs . tricuspidados) e ausência de ganchos nas nadadeiras de exemplares machos ( vs . presença nas nadadeiras anal e ventral) distinguem A . obscurus de A . laticeps . Estas espécies diferem das demais incluídas no complexo por caracteres merísticos e morfométricos e padrão de colorido. É fornecida uma chave de identificação para as espécies do gênero ocorrentes no sistema da laguna dos Patos.


Introduction
included both Astyanax obscurus (Hensel) and A. laticeps (Cope) in the A. scabripinnis species complex, a group previously proposed by Moreira-Filho & Bertollo (1991), and further discussed by Bertaco & Malabarba (2001). According to Bertaco & Lucena (2006), the species of the A. scabripinnis complex possess the deepest and most robust body area close to the middle of the length of the pectoral fins, a robust head, snout short and abrupt, body depth smaller than 41% of SL (mean 30-33% of SL), reduced number of branched anal-fin rays (13-23, usually 17-18, rarely 22 or 23), presence of one or two humeral spots, and a dark midlateral body stripe extending to the tip of the middle caudal-fin rays. Hensel (1870) described Tetragonopterus obscurus as a new species, based on three specimens from rio "Cadea" (Cadeia) above a large waterfall, in Rio Grande do Sul State. Soon after, Steindachner (1876) listed T. obscurus as a junior synonym of T. fasciatus Cuvier, 1819. Afterward, Eigenmann (1910) transferred most species of Tetragonopterus to the genus Astyanax and listed T. obscurus as a junior synonym of Astyanax bimaculatus (Linnaeus, 1758). Later, Malabarba (1989) listed T. obscurus as a synonym of "Astyanax aff. fasciatus (Cuvier)". Lima et al. (2003) listed A. obscurus as a valid species, without any comment. Recently, Melo & Buckup (2006) examined the types of A. obscurus and confirmed the proposition of Lima et al. (2003).
Tetragonopterus laticeps was described by Cope (1894) from Rio Grande do Sul State. Subsequently, Eigenmann (1910Eigenmann ( , 1921 listed Astyanax laticeps as a subspecies of A. scabripinnis (Jenyns, 1842), and in 1927, proposed five subspecies for A. scabripinnis, including A. scabripinnis laticeps. Later, Malabarba (1989) maintained the taxonomic status of this species as previously designated by Eigenmann. Astyanax laticeps was considered for decades as a subspecies of A. scabripinnis. However, Bertaco & Malabarba (2001) examined the types of A. laticeps and concluded that it is a valid species, a proposition subsequently followed by Lima et al. (2003) and Rosa et al. (2009). Both species were originally described from the Atlantic coastal drainages of Rio Grande do Sul State, Brazil. Astyanax obscurus from rio Cadeia (rio Jacuí drainage) and A. laticeps have their type locality restricted to the laguna dos Patos system by Malabarba (1989). The original descriptions are imprecise, and their diagnostic characters were never properly investigated. More recently, A. laticeps was recorded from a limited portion of laguna dos Patos system by Azpelicueta & Loureiro (2009). Herein, we redescribe A. obscurus and A. laticeps based on a new and large series of specimens.

Material and Methods
The specimens examined are deposited in the Academy of Natural Science, Philadelphia (ANSP), Museo Argentino de Ciencias Naturales Bernardino Rivadavia, Buenos Aires Counts were taken as described by Fink & Weitzman (1974), with the exception of the number of scale rows below the lateral line, which followed Bertaco & Lucena (2006). Counts of vertebrae, supraneurals, gill-rakers on the first arch, teeth, and procurrent caudal-fin ray counts were taken from cleared and stained (c&s) specimens prepared according to Taylor & van Dyke (1985). Tooth counts were also taken in all specimens included in the tables. Vertebral counts included the four vertebrae of the Weberian apparatus, and the terminal centrum counted as a single element. In the descriptions, an asterisk indicates counts of the type series. Lower and upper jaws of c&s specimens were prepared for SEM (scanning electron microscopy) in the Centro de Microscopia e Microanálises, CEMM (PUCRS), and Centro de Microscopia Eletrônica, CME (UFRGS).
Measurements were taken point to point with a caliper on the left side of specimens whenever possible. Measurements are expressed as percentage of standard length (SL) except for subunits of the head, which are recorded as percents of head length (HL). In the material examined, the catalog number is followed by the total number of the lot and, in parentheses, the number of specimens counted and measured and respective length range. In cases without parentheses, the range length corresponds to the smallest and largest specimens of the lot.
The geographic variation of A. laticeps was investigated considering the following main drainages (listed from south to north): laguna dos Patos system, rio Tramandaí drainage, rio Mampituba, rio Tubarão, rio Itajaí-Açu, rio Itapocu, and rio Cubatão Norte (Santa Catarina). The samples were analyzed in each one and the sex determined. As no difference was found, the specimens were combined in respective drainages. The specimens from rio Uruguay drainage, coastal drainages from Uruguay, and coastal rivers of Paraná State were not considered for multivariate analysis due to their reduced number and/or size of specimens.
The analysis of geographic variation of the counts of A. laticeps was performed with SigmaPlot 2.0 for Windows 95 and SigmaStat 2.0 for Windows 95. The measures were performed by Principal Component Analysis (PCA) discussed by Bookstein et al. (1985:101), and using the software PAST 1.8 (Hammer & Harper, 2003). Basic descriptive statistics and statistical tests for meristic data follow Weitzman & Malabarba (1999). ( Description. Morphometric data summarized in Table 1. Body compressed and elongate, with greatest body depth anterior to dorsal-fin origin. Dorsal profile of head straight or slightly convex from posterior nostril to tip of supraoccipital spine. Dorsal body profile convex from tip of supraocciptal spine to base of last dorsal-fin ray; straight from latter point to adipose fin origin. Ventral profile of body slightly convex from vertical through posterior nostril to pectoral fin origin, nearly straight to anal-fin origin, and posterodorsally slanted along anal fin base. Caudal peduncle elongate, nearly straight to slightly concave in the dorsal and ventral margins. Snout rounded from margin of upper lip to vertical through anterior nostrils. Head small. Mouth terminal, lower jaw slightly longer than upper jaw. Maxilla extending posteriorly to vertical through anterior margin of orbit, slightly curved, aligned at angle of approximately 45 degrees to longitudinal body axis. Maxilla slightly widened anteroposteriorly.

Astyanax obscurus
Two tooth rows in premaxilla; outer row with three to five (4-5*) pentacuspid teeth, central cusp longer; inner row with  five teeth, gradually decreasing in length from the first to fourth, last distinctly smaller, with five to seven cusps; central cusp twice as long and broad as others cusps. Maxilla with one* (rarely two) teeth, one to three cusps, with central cusp longer. Three to four anteriormost dentary teeth larger, with five or seven cusps, followed by one medium-sized teeth with three to five cusps, and four or six teeth with one to three cusps. Central cusp in all teeth two to three times as long and broad as other cusps. All cusp tips slightly curved posteriorly towards inside mouth (Fig. 3).
Dorsal-fin rays ii,9 (ii,9-10*, one specimen with ii,11, n = 28); first unbranched ray approximately half length of second ray. Distal margin of dorsal fin nearly straight to somewhat convex. Dorsal fin origin approximately at middle of SL.

Color in alcohol.
Dorsal and dorsolateral portions of head and body dark brown. Dark chromatophores scattered on lateral portion of head, more densely concentrated on snout and anterior border of eye. Dorsal portion of body densely pigmented in larger specimens. Body with black, midlateral stripe extending from the second humeral spot to caudal fin base; faint dark pigmentation present on middle caudal-fin rays. Midlateral body stripe expanded dorsally and ventrally to caudal fin base, forming small caudal spot. Two humeral spots. Anterior one, conspicuous, vertically elongate with superior portion wider, located over third to fifth vertical series of scales, extending over 2 to 3 horizontal series of scales above lateral line; inferior portion of spot narrow (1 to 2 scales pigmented), extending over 1 to 2 horizontal series of scales below lateral line. Second humeral spot large, occasionally faint, not ventrally surpassing lateral line,  extending over 3 horizontal series and 2 to 3 vertical series of scales (Figs. 2 and 4a). Region between spots pale. In specimens smaller than 28.0 mm SL superior portion of first humeral spot lightly larger than inferior portion, second humeral spot can be absent. Fins with dispersed dark chromatophores. Specimens a few days fixed in formalin (MCP 40000) with overall body coloration greenish and all fins red and yellowish pigmented.
Sexual dimorphism. Secondary sexual characters were not found on examined specimens. Testes were observed in one dissected c&s specimen (MCP 26125, 67.7 mm SL). Gill glands (Burns & Weitzman, 1996) were not found on first gill arch in all specimens.  Table 1. Morphometric data of Astyanax obscurus: syntypes (ZMB 7478, n = 3), and non-type specimens from rio Cadeia, tributary of rio Caí (A, n = 21), and from rio das Antas, rio Jacuí drainage (B, n = 33); and Astyanax laticeps: holotype (ANSP 21852), and paratypes (ANSP 21743, n = 4), Rio Grande do Sul, Brazil.   Tables 1-2. Body compressed and elongate; greatest body depth usually anterior to dorsal-fin origin. Dorsal profile of head between vertical through posterior nostril and tip of supraoccipital spine straight. Dorsal body convex profile from tip of supraocciptal spine to base of last dorsal-fin ray; straight from latter point to adipose fin origin. Ventral profile of body slightly convex from vertical through posterior nostril to pelvic fin insertion, nearly straight from that point to anal fin origin, and posterodorsally slanted along anal fin base. Caudal peduncle elongate, nearly straight to slightly concave in the dorsal and ventral margins. Snout rounded from margin of upper lip to vertical through anterior nostrils. Head small. Mouth terminal. Maxilla extending posteriorly to vertical through anterior margin of orbit, slightly curved, and aligned at an angle of approximately 45 degrees relative to longitudinal body axis. Maxilla slightly widened posteriorly.

Description. Morphometric data summarized in
Two tooth rows in premaxilla; outer row with four to five* tricuspid teeth with central cusp longer; rarely, the most lateral teeth with two additional very small cusps. Five teeth in inner premaxillary row; teeth gradually decreasing in length from first to fourth teeth with last tooth distinctly smaller: teeth with three to five cusps with central cusp twice as long and broad as other cusps; rarely, two most lateral cusps present, but visible only in c&s specimens. Maxilla with one to three (two to three*) teeth with one to three cusps, with central cusp longest. Four anteriormost dentary teeth larger, with five cusps, followed by five to seven small teeth with one to three cusps. Central cusp in all teeth two to three times as long and broad as other cusps. All cusp tips slightly curved posteriorly towards inside mouth (Fig. 8).

Color in alcohol.
Dorsal and dorsolateral portions of head and body dark brown. Dark chromatophores scattered on lateral portion of head, more densely concentrated on snout and anterior border of eye. Dorsal portion of body densely pigmented in larger specimens. Scales midlateral and below the lateral line bordered with dark brown chromatophores forming reticulate pattern. Body with black, midlateral stripe extending from the humeral spot to caudal fin base; faint dark pigmentation present on middle caudal-fin rays. Midlateral body stripe expanded dorsally and ventrally to caudal fin base, forming small caudal spot. Humeral spot, conspicuous, horizontally elongate located over 3 to 5 vertical series of scales and extending over 2 horizontal series of scales above of the lateral line; with a narrow anteroventral downward extension past lateral line and extending over one, rarely two, vertical series of scales (Figs. 4b and 7). Humeral spot can vary in shape with upper portion almost rounded or slightly wider than lower portion in small specimens around 40.0 mm SL (MCP 40807,and MUCP 940) and occasionally in large specimens (53.1 mm SL, MUCP 3250, and 66.8 mm SL, MCP 34533). Fins with dispersed dark chromatophores. Specimens a few days fixed in formalin (MCP 33312), all fins red-orange pigmented.
Sexual dimorphism. Males with thick, well-developed retrorse bony hooks on pelvic-fin and anal-fin rays. One paired bony hook per lepidotrichia in the last unbranched anal-fin ray and first to eighth anal-fin branched rays on middle and distal portions of rays. One paired bony hook per lepidotrichia on entire first to sixth pelvic-fin branched rays. Anal fin profile slightly concave to straight in males, concave in females. In males up to 60 mm SL, tip of pelvic fin reaches or slightly passes the anal fin origin, and tip of pectoral fin extends beyond the pelvic fin origin. In females, both fins do not reach the anal and pelvic fin origins, respectively. Gill glands   (Burns & Weitzman, 1996) were not found macroscopically on first gill arch in either males or females.
Distribution. Astyanax laticeps is known from the coastal drainages of Uruguay (arroyo Don Carlos) and Brazil to south of the Paraná State (rio Cubatão draining to baía de Guaratuba), and rio Uruguay drainage (Fig. 5). This finding substantially extends the distribution of the species reported by Azpelicueta & Loureiro (2009).
Geographic variation. All meristic characters analyzed among A. laticeps populations showed wide overlap, except for three counts (lateral line and caudal peduncle scales, and branched anal-fin rays). Kruskal-Wallis nonparametric multiple comparisons indicate significant differences between population samples. Dunn's method shows differences in the number of lateral line scales between the rio Cubatão Norte and rio Itapocu populations and other populations, except that from the rio Tramandaí drainage (Fig. 9). The same count was not significantly different between the rio Cubatão and rio Itapocu populations. In the counts of scales around the caudal peduncle, the rio Cubatão Norte population was significantly different from all other populations, except the rio Itapocu (Fig. 10). The same pattern occurred in the number of branched anal-fin rays between the rio Cubatão Norte and other populations, except the rio Itapocu. This count was significantly different between the rio Itapocu and other populations, except the rio Mampituba (Fig. 11). Significant differences were found in the number of lateral line scales between the rio Mampituba and rio Tubarão populations and the others, except the rio Itajaí and laguna dos Patos system.
The morphometric data analyzed by percentages and PCA did not reveal at all the discriminate populations (Figs. 12-14, Table 3). The Mampituba population can be partially discriminated from those of the laguna dos Patos, Tramandaí, and Itapocu drainages. The variables that most influenced the discrimination on the third axis (explaining 0.89511% of the variance) were interorbital width and caudal peduncle length on the positive side and anal fin base length and snout length on the negative side (Fig. 12). When only females were analyzed, the populations of the Tramandaí and Mampituba drainages can be partially discriminated from each other. The most influenced measurement was the caudal peduncle length and interorbital width on the negative side on the second axis (explaining 0.7729% of the variance) and the anal fin base length and snout length on the positive side (Fig.  13). The PCA for males showed that the populations Tramandaí  and Mampituba, and Itajaí and Cubatão Norte drainages are discriminated from each other on the second axis (explaining 2.4185% of the variance) (Fig. 14). Pectoral and pelvic fin lengths were the variables that were influenced most on the negative side, and snout length and body depth on the positive side. Despite the above-mentioned differences, there is overlapping of ranges of meristic and percentage data. In the meristic data analyzed, the Itapocu and Cubatão Norte populations showed higher median values compared to the other populations, suggesting that these populations may be distinct. However, there is overlapping of the extreme values of these characters when compared with southern populations. Besides, no diagnostic characters were found to discriminate the Itapocu and Cubatão Norte populations from the others. The few differences in PCA did not show any rational geographic pattern in the entire distribution area of the species. Therefore, we considered all populations analyzed as Astyanax laticeps.
Among the discerning taxonomic studies, including species widely distributed in the coastal rivers of south and southeast Brazil, only Mimagoniates microlepis (Steindachner) showed a distribution similar to that of A. laticeps (Weitzman et al., 1988;Menezes & Weitzman, 1990. Weitzman et al. (1988) found divergent live color patterns between some isolated populations of M. microlepis, and related this divergence to the possible initial process of speciation. In the present study, the tendency toward higher median values for the populations from Itapocu and Cubatão drainages could be indicative of the same early process.

Discussion
Astyanax obscurus and A. laticeps have the characters and general body morphology of the species of the A. scabripinnis species complex which was discussed by Moreira-Filho & Bertollo (1991) and Bertaco & Malabarba (2001) and later tentatively delimited by Bertaco & Lucena (2006).
The species name A. scabripinnis has been largely used to refer to populations of Astyanax from the rio Paraná basin and coastal river drainages of south and southeast Brazil that possess similar features described for the A. scabripinnis species complex (e.g., Castro & Vari, 2004;Villela et al., 2002). According to our results, most of the citations of A. scabripinnis for the rio Uruguay drainage (Rio Grande do Sul State) and costal river drainages of the States of Santa Catarina and Paraná (rio Cubatão), must be referred to as A. laticeps. Recently, Miquelarena & Menni (2005) recognized A. scabripinnis in their key for Astyanax species from Argentina. According to these authors, A. scabripinnis occurs in the Paraná and Uruguay River drainages in Argentina. The features of the humeral spot mentioned in the key indicate that the specimens examined by these authors are not A. laticeps or A. scabripinnis. According to Melo (2001) and Bertaco & Lucena (2006), the occurrence of A. scabripinnis is restricted to Rio de Janeiro State, and the species may be extinct. A review of populations of Astyanax from the rio Paraná basin is necessary to explain the citation of A. scabripinnis for this basin.
Astyanax obscurus remained for one hundred and thirty years as a junior synonym of other Astyanax species and was considered only recently as a valid species (Lima et al., 2003;Bertaco & Lucena, 2006;Melo & Buckup, 2006) but without a taxonomic analysis. According to our results, the conclusion of these authors is confirmed here. Bertaco & Lucena (2006) included fifteen species in the A. scabripinnis species complex, distributed throughout south and southeast Brazil (drainages of the rios São Francisco, Itapemirim, Paraíba do Sul, Paraná, Uruguay and Tramandaí, and laguna dos Patos). Although A. obscurus is sympatric with A. laticeps, some differences between their habitats can be observed. Astyanax obscurus inhabits mainly headwater environments of the laguna dos Patos system. On the other hand, A. laticeps seems to exploit a larger diversity of niches and may also occur in the middle and lower courses of rivers, consequently showing a wider distribution, from coastal rivers in Uruguay to southern Paraná State in Brazil, and in the rio Uruguay drainage.
Haluch & Abilhoa (2005) mentioned two other populations with a humeral spot similar to that of A. laticeps, Astyanax sp. D from rio Iguaçu basin and A. aff. scabripinnis from rio Morato, coastal drainage of Paraná State. Astyanax laticeps differs from Astyanax sp. D by the absence of bony hooks on the pectoral fin of males, and from A. aff. scabripinnis by the number of gill rakers in the lower branch of the first arch (10 vs. 11-16).
Therefore, with the increased number of Astyanax species that occur in the laguna dos Patos system, along with the difficulty that biologists have in the identification of these species, we find it opportune to present an identification key for these species. Some species included in the key may also occur in the drainages adjacent to laguna dos Patos system, namely A. brachypterygium (upper rio Uruguay), A. cremnobates (rio Tramandaí system), A. eigenmanniorum, A. sp. aff. fasciatus (sensu Melo & Buckup, 2006), and A. jacuhiensis. The last three species are the object of taxonomic studies by the authors to determine their identity and true distribution.