Mimagoniates rheocharis, Menezes & Weitzman, 1990

Menezes, Naéreio A. & Weitzman, Stanley H., 1990, TWO NEW SPECIES OF MIMAGONIA TES (TELEOSTEI: CHARACIDAE: GLANDULOCAUDINAE), THEIR PHYLOGENY AND BIOGEOGRAPHY AND A KEY TO THE GLANDULOCAUDIN FISHES OF BRAZIL AND PARAGUAY, Proc. Biol. Soc. Wash. 103 (2), pp. 380-426 : 399-422

Mimagoniates rheocharis , new species

Figs. 17-28 , Table 2

Species B. — Weitzman et al., 1988:figs. 6, 10, 23 [phylogeny and biogeography].

Holotype. — MZUSP 40278 , male, SL 47.3 mm, Brazil, Santa Catarina, Municipio de Praia Grande, rio Faxinalzinho at Mãe dos Homens , near Praia Grande , approximately 29°20' S, 14°40' W, 9 Jun 1985; C. A. S. de Lucena, R. E. Reis, and L. R. Malabarba.

Paratypes.-Following 2 lots of immatures to adults collected with holotype: MCP 13616 , spms. 3, SL 24.3-40.8 mm; USNM 279878 , spms. 3, SL 33.3-47.8 mm. Following 3 lots of immature to adult paratypes all collected 7 Jun 1985 by C. A. S. de Lucena and party: Brazil, Santa Catarina, Municipio de Nova Veneza, rio Jordäo at Jordão Alto , approximately 28 °36’ S, 49°28 ' W, MCP 13617 , spms. 83, SL 21.9- 39.0 mm; MZUSP 40279, spms. 81, SL 22.4- 42.8 mm; USNM 270879, spms. 89, SL 22.8-49.0 mm. Following 3 lots of immature to adults all collected by R. M. C. Castro 13 Feb 1988: Brazil, Santa Catarina, Município de Praia Grande, tributary of rio Grande approximately 29°20 ’ S, 49°40 ’ W, MZUSP 40280, spms. 29, SL 13.5-38.4 mm; USNM 306339, spms. 18, SL 17.0- 40.0; spms. 2 to R. M. C. Castro, SL 28.0- 31.30 mm. MZUSP 40281, SL 41.8 mm (mature male), Brazil, Rio Grande do Sul, Municipio de Osório, arroio das Pedras, approximately 29°52' S, 50° 19 'W, ll Feb 1988, R. M. C. Castro. MCP 13613 spms. 7, SL 27.5-31.3 mm, Brazil, Município de Osório, arroio Água Parada, tributary to rio Maquiné, in Maquiné, approximately 29 °40' S, 50° 11 ’ W, 10 Oct 1989, R. E. Reis, S. O. Kullander, L. R. Malabarba and J. Pozzi. MCP 10806, SL 39.5 mm (mature male), Brazil, Rio Grande do Sul, Municipio de Torres, tributary of rio Três Furquilhas, Chapéu, approximately 29°19 ’ S, 49°44' W, 25 May 1986, C. A. S. Lucena, R. E. Reis, and L. R. Malabarba.

Diagnosis. — Mimagoniates rheocharis may be separated from all other species of Mimagoniates by use of the key provided above and the characters described and discussed below. Minıagoniates rheocharis is sympatric with and apparently at least partly syntopic with M. microlepis in south coastal Santa Catarina. It is also sympatric and syntopic with M microlepis in the rio Maquiné drainage. The southern distribution of M rheocharis (rio Maquiné drainage and streams immediately surrounding Osório in Rio Grande do Sul) is adjacent to the northeastern limit (rio Gravatai drainage) of M. inequalis where these two species abut and appear to be allopatric (see the Biogeography section and phylogenetic discussion in the Discussion section below). The possible relationships of M rheocharis to M. inequalis or M. microlepis is uncertain and perhaps complex but the following series of characters distinguish M rheocharis from one or the other of these species or, in some cases, both of them. Because the comparisons are somewhat complex, for clarity the two compared species are treated separately.

Mimagoniates rheocharis may be distinguished from M microlepis , by the following characters (see also the statistical analyses of overlapping characters under Statistical Comparisons below). Scale rows around caudal peduncle 19-23 (15-18 for M. microlepis ); scale rows between dorsal-fin and anal-fin origins 17-22, rarely 17 (14- 17, rarely 17 for M microlepis ); total lateral series scale count shows broad overlap between both species, 41-48 (42-49 for M microlepis ) but in at least one population comparison of M. rheocharis and M. microlepis showed a statistically significant difference (see the statistical comparisons below); branched dorsal-fin ray count 8-12, usually 9-10, rarely 8 (7-9, rarely 9 for M microlepis ); branched anal-fin ray count 23- 29, usually 24-26 and rarely 28 or 29 (26- 33, usually 28-31 and rarely 26 or 27 for M microlepis ); total vertebral count 35-40, usually 36-38, rarely 39 or 40 (38-41, usually 39-40, rarely 38 in M. microlepis ); tenth and eleventh principal caudal-fin rays of adult males with small, little developed hooks posterior to caudal organ but with sturdy hooks on ray 1 1 along ventral border of expanded ray segments of caudal organ, Fig. 24 (adult male M. microlepis with well-developed hooks on caudal-fin rays 7-12 and no sturdy hooks on principal ray ll along ventral border of expanded ray segments of caudal organ, note, some populations with few or no hooks on rays 7 and/ or 12); preserved and live colors in part different between both species, but variation in live colors among various population samples of M microlepis make comparisons difficult.

A few life color characters appear consistently different in fully mature males: pelvic-fin rays and membranes of adult males distally white (adult male M microlepis with yellow and/ or black pigment of pelvic fins continuous to edge of fin where fin bordered by narrow band of white); anal fin posterior to anterior lobe bordered by broad band of deep yellow pigment, very little to no black pigment on fin (adult male M microlepis with posterior portion of anal fin ventrally bordered by narrow band of black pigment, none or very little yellow pigment).

As treated below in the Discussion, there are some, perhaps plesiomorphic. body shape similarities shared between M. rheocharis and M inequalis . Furthermore, M rheoclzarís appears to have the caudal peduncle and body depths for both sexes intermediate between those of M. microlepis and M inequalis ; see Figs. 17 and 27 . Certain aspects of the caudal pump morphology of M. rheocharir could also be interpreted as intermediate. This suggests the possibility that M. rheacharis might be closely related to M. inequalís rather than to M. microlepis or even derived by introgression from ancestral sympatric populations of M microlepis and M inequalis in a manner suggested by Hubbs (1955: 19) for freshwater fishes in North America. However, not all counts or body proportions of M rheocharis are intermediate between the two species and we question the intermediacy of M rheocharis based on tentative polarity and parsimony analyses of the available data (see Discussion below). In any case, the similarities of many features, especially of immature or just maturing specimens, shared between M rheocharis and M inequalis makes it a practical necessity to describe the differences in some detail between the population samples at hand of these two species. The phylogenetic polarities of most of these characters are relatively uncertain for those species possessing them. Therefore their use as synapomorphies suggesting relationships to and/ or among species of Mimagoniates that may share them remains to be more fully investigated. Tentatively, many of these characters appear plesiomorphic for M. inequalis when the latter is compared to M rheocharis or M. microlepis .

For convenience the following same characters as those treated above to separate M rheocharis and M. mícrolepis are considered for comparison between M. rheocharis and M. inequalis even though not all these characters are diagnostic for separation of the latter two species (see also the section below on Statistical Comparisons). Scale rows around caudal peduncle = 19-23 for M rheocharis (16-19 for M. inequalis ) (both species commonly with a count of 19); scale rows between dorsal-fin and anal-fin origins = 17-22 (16-18 for M. inequalis ); lateral series scale count = 41-48, rarely 41 (36- 41 for M. inequalis , 41 being uncommon); branched dorsal-fin ray count 8-12, usually 9 or 10, only occasionally 8 (8-9, rarely 9 for M inequalis ); branched anal-fin ray count 23-29, usually 24-26, rarely 28 or 29 (24-30, rarely 24 for M inequalis ), these counts significantly different statistically between some population samples but not in others (see statistical analyses below); total vertebral count 35-40, rarely 39-40 (36-39 for M. inequalis ); principal caudal-fin rays 10 and ll of adult males with small, little developed hooks posterior to caudal-fin organ, see Fig. 24 (adult male of M inequalis without hooks on caudal-fin rays 10 and 11); preserved color patterns essentially indistinguishable between the two species and we have not been able to study the live color patterns of adult males where differences between the species would most likely be present.

Description-Table 2 presents morphometrics of the holotype and paratypes. Except where noted, the entire description refers to the population sample from near Praia Grande, southern Santa Catarina. These collections were treated statistically as one population sample in an attempt to represent the species as a whole from this area. Counts and ratios of measurements for other population samples taken from a tributary of the rio Grande are given only when they differ from those from near Praia Grande.

Body compressed, relatively deep, especially near dorsal-fin origin; body deepest at vertical line through anal-fin origin. Predorsal body profile relatively arched in adult males, less so in adult females and immatures which have predorsal profile gently convex to tip of snout. Body profile elevated at dorsal-fin origin, still strongly arched in males, less so in females and juveniles. Dorsal body profile nearly straight along dorsal-fin base to adipose fin. Body profile posterior to adipose fin somewhat concave dorsal to caudal peduncle, ending at origin of procurrent caudal-fin rays. Dorsal-fin origin nearer to caudal-fin base than to snout tip. Ventral body profile strongly convex in adult males from anterior tip of lower jaw to origin of pelvic fins, less strongly convex in females and juveniles. Belly profile in adult males slightly concave to anal-fin origin, straight or nearly so in females and juveniles. Body profile along anal-fin base in males slightly concave at anterior base in region of anterior lobe of anal fin; straight along base of remainder of fin in males and along entire anal-fin base in females and juveniles. Ventral profile of caudal peduncle convex in adult males when anterior 5 or 6 strongly developed procurrent ventral caudal-fin rays are included in that profile; slightly concave or nearly straight in females and juveniles.

Head and snout of moderate size in proportion to body length. Lower jaw protruding, slightly anterior to upper jaw. Lower jaw of adult males thick and heavy compared to that of females and juveniles. Mouth angled posteroventrally from anterior tip of snout to posterior part of mandibular joint. Maxilla long, extending to a point ventral to a horizontal line drawn from ventral border of eye in juveniles and adults of both sexes. Maxilla extends posteriorly to a point anterior to vertical line drawn through anterior border of pupil of eye.

Dorsal-fin rays ii, 9 (unhranched rays ii in all specimens, branched rays x = 8.9, range = 8-10, n = 56); posterior ray not split to its base and counted as l. Adipose fin present, slender. Anal-fin rays iv, 26 (iv or v, usually iv, ic = 26.8 for branched rays, range = 25-29, n = 56); posterior ray split to its base and counted as l. Anal fin with moderately developed lobe anteriorly ( Figs. 18 , 19 ). Lobe includes fourth or fifth undivided ray and first 3 divided rays. Anal fin of sexually mature males with bilateral hooks, 1 on each side, on anterior unbranched ray iv or v, whichever occurs just before branched rays ( Fig. 21 ). Usually anterior 7 branched fin rays with bilateral hooks, l set for each ray. Pectoral-fin rays i, lo (unbranched ray i in all specimens, branched rays x = 10.1, range 10- 1 1, a n = 56). Posterior tips oflongest pectoral-fin rays extend posteriorly beyond origin of pelvic fins; fins of about equal extent in both sexes. Pectoral-fin rays without hooks. Pelvic fin rays 8 in all specimens. Pelvic fin with anterior (first) ray branched once, branches remaining close together and entire ray tapering as in most characiforms that have anterior ray unbranched ( Fig. 22 ). Sexually mature, large adult males with over 400 hooks on each pelvic fin.

Principal caudal-fin ray count 10/9 in all specimens, (n = 56). Fin rays modified in association with caudal pheromone pump as in Figs. 23 and 24 . Fin rays modified more like those in M microlepis than any other species of Mimaganiates ( Fig. 5 ). Caudal-fin rays 10 and ll with small bony hooks. Ventral borders of 4 anterior expanded ray segments of ray l l. which form anterior external wall of dorsal portion of pump chamber, with about 4 short but large hooks, 2 middle hooks often bicornate, others 1 hooked.

Scales cycloid, with few radii along posterior border. Terminal scale of modified caudal-fin series without exaggerated radii ( Fig. 24b ).

Lateral line incomplete, perforated scales 7 (x = 6.7, range 5-9, n = 48). Lateral series scales 45 (x = 44.3. range = 41-48, n = 48). Predorsal scales = 22 (x = 22.3, range = 21-24, n = 46). Scale rows between dorsal-fin and anal-fin origins 19 (x = 18.8, range = 17-21, I’ n! = 55). Scale rows around caudal peduncle 20 (x = 19.7, range = 19-22, n = 46).

Premaxillary teeth in 2 distinct rows, Fig. 25 . Larger and smaller teeth tricuspid in all large specimens, sometimes in smaller specimens small teeth bicuspid or conical. Outer row teeth 5 (x = 6.2, range 5-8. n = 55). Inner row teeth few, 3 (x = 2.1 J, range = 1- 3, n = 55). Outer and inner row premaxillary teeth somewhat compressed compared to most “ tetragonopterine ” characid teeth which are often circular in cross section. Maxillary teeth 8 (x = 5.4. range = 3-8, larger specimens usually with higher counts, n = 55). All maxillary teeth usually tricuspid in large sepcimens; small specimens with posterior maxillary teeth often conical. Dentary with 4 large tricuspid teeth in all specimens, smaller posterior teeth 12 (x = 8.4, range = 5-12, almost always greater number of teeth in largest specimens, anterior small maxillary teeth tricuspid, posterior ones conic, n = 54). Maxillary and dentary teeth shaped much like premaxillary teeth described above. No significant difierence in tooth number found between males and females.

Vertebrae 38 (X = 38, range = 37-40, n = 107). Dorsal limb gill rakers 7 (x = 6.6, range = 6-8, n = 54); ventral limb gill rakers 12 (x = 11.7, range = 11-13, n = 54). Branchiostegal rays 4, in 3 cleared and stained specimens, 3 rays originating on anterior ceratohyal and 1 ray from posterior ceratohyal.

'

Color in alcohol.- See Figs. 18-20 for preserved color pattern of males and females. Body pale to medium brown, pale yellowish brown ventrally, much darker dorsally. Lateral body stripe diffuse in both sexes, especially anteriorly. Stripe extending from vertical humeral spot posteriorly to caudal fin and onto dorsal region of ventral caudal-fin lobe and small part of dorsal caudal-fin lobe. Lateral stripe extends over all caudal gland structures, including those derived from dorsal caudal-fin lobe such as modified caudal squamation. Stripe especially dark on principal rays 10, 11 and 12, less so on ray 13. Humeral spot vertically elongate, especially in sexually mature males. Remainder of caudal fin dusky because of scattered dark chromatophores. Dorsal border of first principal caudal-fin ray and ventral border of nineteenth principal caudal-fin ray black. Utmost dorsal body surface black, forming a narrow stripe extending from supraoccipital region to base of dorsal procurrent rays of caudal fin. Remainder of dorsal body surface ventral to lateral body stripe pale brown.

Pectoral, pelvic, dorsal and anal fins dusky with scattered dark chromatophores along fin rays and membranes. Anal fin with a dark elongate stripe running length of fin. Stripe width about one-fourth height of anal fin anteriorly and about one-halffin’s height posteriorly. Stripe of nearly uniform width throughout its length, but fin height changes, being considerably shorter posteriorly. In sexually mature males stripe considerably darker anteriorly, especially dorsal to anterior anal-fin lobe which appears relatively hyaline. Dorsal-fin with a horizontal dark stripe in adult males and females extending posteriorly from about mid-length of anterior elongate undivided ray to posterior tips of terminal two dorsal-fin rays. Stripe usually narrow, less than one-eighth height of dorsal fin at latter’s longest measurement. Width and density of stripe variable depending on sex and sexual maturity. Preserved males sometimes with posterior portion of stripe diffuse. Adipose fin dusky with scattered dark chromatophores.

Head dark brown around mouth and on dorsal surface of snout, between eyes, dorsum of cranium and nape. Mental area of lower jaw dark brown. Head area posterior to circumorbitals and extending ventrally from parietal region, across dorsal opercular region dark brown. Dark area continues ventrally across posterior region of opercular bone to just reach interopercular bone; looks like an anteriorly misplaced humeral spot. Iris dorsal to pupil dark brovm, most of remainder of iris silvery. Circumorbitals silvery if guanine preserved, pale yellowish brown if guanine destroyed by formalin. Dark brown chromatophores scattered evenly through circumorhital area. Anterior area of opercle, all of preopercle, and branchiostegal rays silvery or pale brown, without much dark brown pigment.

Color in life. —Life colors described here taken from a 35 mm color slide made by Ricardo M. C. Castro of an adult male 41.8 mm SL (MZUSP 4028 1). See also black and white photograph of this specimen in preservative, Fig. 20 . Specimen photographed just after capture from a clear water stream surrounded by vegetation. Site located immediately north of city of Osório, Rio Grande do Sul. Sides of body pale silvery blue. Broad lateral body stripe somewhat deeper silvery blue from humeral spot to caudal peduncle termination. Just dorsal to silvery blue color of body sides, back with a narrow dark brown line extending from parietal region of head to just ventral to adipose fin. Lateral portion of back between narrow brown line and dorsomedian narrow dark brown line extending across dorsalmost portion of back, a brovwiish yellow green color. Dorsal region of caudal peduncle nearly yellow. Ventral abdominal area, most of lower jaw, ventral opercular area, branchiostegal rays and their membranes silvery white. Dark pigment of head similar to that described for preserved specimens except that dorsal region of opercle appears silvery blue. Dorsal caudal-fin lobe and principal caudal-fin rays 14- 16 on ventral caudal-fin lobe bright yellow, except for black proximal half of ray 14. This black pigment continuous with black pigment surrounding structures of caudal pheromone organ. Rays 17-19 (ventral caudal-fin lobe) hyaline or nearly hyaline except ray 17 which is somewhat yellow. Remainder of black pigment of caudal fin as described above for preserved specimens.

Anal fin with distal portion of fin rays posterior to anterior anal-fin lobe lemon yellow, forming a stripe along ventral border of fin. Distal region of anterior anal-fin lobe hyaline to white, proximally bordered by black pigment described above for preserved specimens. This black pigment mixed with yellow. Basal half of anal fin hyaline with some scattered brown chromatophores and a small amount of yellow anteriorly. Posterior portion of narrow horizontal black line of anal fin pale. Approximately distal half of pelvic fin white; a black and yellow band proximal to this; remaining proximal portion of fin hyaline. Distal half of pectoral fins yellow, proximal half with black rays and yellow membranes. Dorsal fin hyaline to white distal and proximal to longitudinal black and brown longitudinal stripe.

Sexual dimorphism. —Females lack a caudal pheromone pump organ and pelvic-fin and anal-fin hooks described above for males. Live color of females is unknown but undoubtedly more subdued than that of sexually mature males. Figure 17 graphically indicates that according to these population samples, males and females of M. rheocharis and M. micralepis respectively show no significant sexual dimorphism in body depth. In an F-ratio test for homogeneity of slopes in an analysis of covariance for body depth on SL of M. rheocharis , no significant difference was found for 31 males and 25 females from Praia Grande (MZUSP 40270, 40280, MCP 13616, and USNM 279878, 306339. The above sample included very young as well as fully adult specimens. Juveniles of undetermined sex were entered as females since young females are indistinguishable from juveniles in body depth.

In another sample of M. rheocharis consisting of only adults from rio Jordão (MZUSP 40279, MCP 13617 and USNM 279879), Fig. 25 , the same statistical test for body depth as a function of SL in 2.7 males and 23 females demonstrated no significant difference between slopes but does show a significant difference (F ₀.₀₅,(₁, ₄₇) 18.67, P <0.001) in adjusted means. Both Figs. 25 and 26 indicate considerable sexual dimorphism in body length for M. rheocharis .

=

Figures 27 and 28 provide graphic evidence that there appears to be a divergence in caudal peduncle depth between males and females of M rheacharis as they become sexually mature and increase in length. We have few females over 33.0 mm in SL so the nature of caudal peduncle depth divergence between males and females remains unsampled if females reach the same SL as males. We doubt that happens, because in all the population samples of various species of Mimagoniates examined by us, the males reach a greater adult length than females. Thus, again in our analyses of sexual dimorphism our comparison of males and females of unequal body lengths is probably an expression of reality rather than of missing data for females. In F -ratio tests of an analysis of covariance for caudal peduncle depth on SL, no significant difference was found for the homogeneity of slopes between 30 male specimens and 25 females although a significant difference (F ₀.₀₅, ₍₁,₅ ₂ ₎ = 22.33, P <0.001) in adjusted means was evident; see also Fig. 27 . This population sample is from the specimens reported above from near Praia Grande. In another population sample, the same one reported above regarding body depth as a function of SL from rio Jordão, the slopes of the females versus males were not significantly different in caudal peduncle depth as a function of SL ( Fig. 28 ) but the adjusted means are significantly different (F₀.₀₅,₍ ₁, ₄₇₎ = 18.67, P <0.001) and are similar to those reported above from near Praia Grande rio Grande region, Santa Catarina.

05_

_

Statistical comparisons. —As noted in the Diagnosis above and the Discussion below of M rlzeocharis , comparisons testing hypotheses of significant statistical difierence between M rheocharis and two species, M. mícrolepis and M. inequalis . were necessary because many of their characters overlap and because of the need to examine, in a preliminary way, the possibilities of introgression among these species.

Body depth as a function of SL in an F -ratio test for homogeneity of slopes ( Fig. 17 ) showed no significant differences in an analysis of covariance between 31 males of M. rheocharis from rio Grande region, Santa Catarina (MZUSP 40280, USNM 306339) and 39 males of M microlepis from the area near Paranaguá, Paraná (MZUSP 40288, USNM 306378) but did show a significant difference (F₀.₀₅,₍₁,₆₇₎= 188.05, P <0.000) in adjusted means. Same analysis for 24 and 31 females respectively from same localities also showed no significant difference in slopes but did show such a difference (F₀.₀₅,₍₁,₅ ₃ ₎ = 82.97, P <0.000) in adjusted means.

Caudal peduncle depth as a function of SL in an F-ratio test for homogeneity of slopes ( Fig. 27 ) showed no statistically significant difference in an analysis of covariance between 30 males of M rheocharis from near Praia Grande, rio Grande region. Santa Catarina (MZUSP 40280, USNM 306339) and 40 males of M. microlepis from area near Paranaguá, Paraná (MZUSP 40288, USNM 306378) but did show a significant difference (F ₀.₀₅,₍ ₁, ₆₇₎ = 94.36, P = 0.000) in adjusted means. Same analysis for 26 and 31 females respectively from same localities also showed no significant difierences in slopes but did show such a difference (F ₀.₀₅ ₍₁,₅₄₎ = 106.03, P <0.000) in adjusted means.

Number of scale rows between dorsal-fin origin and anal-fin origin of M rheocharis (17-22) and M. lnicrolepis (14-17) overlaps somewhat but is statistically significantly different. Forty-two specimens of M. rheocharis (x = 19.6, range = 18-22, SD = 0.8211) from rio Jordão (MZUSP 40279, MCP 13617, USNM 270879) and 58 specimens of M. microlepis (x = 15.2, range = 14- 17, SD 0.6702) do not overlap but 53 specimens of M rheocharis (x = 18.8, range = 17-21, SD = 0.8905) from the rio Grande region of Santa Catarina do overlap with the specimens of M microlepis just noted. These last two samples are significantly different (t = 24.959, P = 0.00).

Lateral series scale counts, as noted above broadly overlap but one case of 48 specimens of M rheocharis (x = 44.2, range = 41-48, SD = 1.7704) from rio Jordão (MZUSP 40279, MCP 13617, USNM 270879) and 45 specimens of M micralepis (x = 45.0, range = 42-49, SD = 1.5948) from Paranaguá and rio Nhundiaquara (MZUSP 40281, USNM 249886, 249890, 249894, 249897, 257114, 257115, 257198) showed a significant difference (t = 2.459, P = 0.015).

Branched dorsal-fin ray count higher (8- 12, x = 9.4) for M. rheocharis than any other species in tribe Glandulocaudini (branched dorsal-fin rays almost always 8, rarely 7 or 9) and constitutes an autapomorphy for the species. Two sampled populations differed considerably from one another in this count. Fifty specimens of M. rheocharis (x = 9.8, range = 9-12, SD = 0.6701) from rio Jordão (MZUSP 40279, MCP 13617, USNM 270879 and 56 specimens of this species (x = 8.9, range = 8-10, SD = 0.4615) from rio Grande region of Santa Catarina (MZUSP 40280, USNM 306339) were significantly diiferent (t = 7.901 ›, P = 0.00). Both of these population samples of M. rheocharis differed significantly from that of 70 specimens of M microlepis (x = 8.0, range = 7-9, SD = 0.1703) from area near Paranaguá and rio Nhundiaquara, Paraná (MZUSP 40281, 40288, USNM 249886, 249890, 249894, 24897, 257114, 257115, 257198, 306378). Thus, t = 22.490, P = 0.00 for rio Jordão sample and t = 9.472, P = 0.00 for rio Grande sample of M. rheocharis .

Branched anal-fin ray counts, although overlapping for these two species are statistically different (t = 19.402, P = 0.00) in 50 specimens of M. rheocharis (fc = 25.3, range = 23-27, SD = 1.1073) from rio Jordão (MZUSP 40279, MCP 13617, USNM 270879) and 70 specimens of M. microlepis (x = 29.7, range = 26-33, SD = 1.2471) from near Paranaguá and rio Nhundiaquara at Morretes, Paraná, (MZUSP 40281, 40288, USNM 306378, 249886, 249890, 249894, 249897, 257114, 257115). A second population sample of 56 specimens of M. rheoclıaris (x = 26.8, range = 25-29, SD = 1.3088) from area of rio Grande, Santa Catarina (MZUSP 40278, 40280, MCP 13616, USNM 279878, 306339) also differed significantly in a similar test from the Paranaguá and rio Nhundiaquara samples (t = 12.1510, P = 0.00).

Vertebral counts of M. rheocharis and M. microlepis overlap somewhat, but they are significantly dilferent (t = 12.528, P <0.00): x = 38.0, range = 37-40, SD = 0.6558, n = 87 for M. rheacharis from rio Jordão (MZUSP 40279, USNM 270879) and x = 39.3, range = 38-41 J, SD = 9.6532, n = 70 for M. microlepis from rio Nhundiaquara (MZUSP 40281, USNM 249886, 249890, 249894, 249897, 257114, 257115, and 257198).

In the following statistical comparisons of M. rheocharis and M. inequalis , specimens of the latter species from an arroio near Morungava, Município de Gravataí, Rio Grande do Sul (MAPA l 15), were chosen for comparison because they came from the rio Gravataí system, a drainage adjacent to the streams around Osório which contain M rheocharis . See also in the Discussion below comparison of our few specimens of M. rheocharis from the adjacent streams around Osório and the rio Maquiné just to the north of Osório.

Body depth as a function of SL in an F -ratio test for homogeneity of slopes ( Fig. 17 ) showed no significant difference in an analysis of covariance between 31 males of M. rheocharis from rio Grande region, Santa Catarina (MZUSP 40280, USNM 306339) and 26 males of M inequalis from Morungava, Rio Grande do Sul (MAPA 1 15) but did show a difference (F₀.₀₅,₍₁,₅₄₎ = 50.03, P <0.000) in adjusted means. Same analysis for 25 females and 14 females of each species respectively from same localities showed a significant difference (F₀. ₀₅, ₍₁, ₃₆ = 42.41, P <0.000) in body depth on ₎ SL only for adjusted means.

Caudal peduncle depth as a function of SL in an F-ratio test for homogeneity of slopes ( Fig. 27 ) showed no statistically significant difference in an analysis of covariance between 30 males of M. rheocharis from rio Grande, Santa Catarina (MZUSP 40280, USNM 306339) and 26 males of M. inequalis from Morungava, Rio Grande do Sul (MAPA 1 15) but did show a significant difference (F₀.₀₅ ₍₁.₅₃) = 43.95, P <0.000) in adjusted means. Some test for 26 females and 14 females respectively from same localities showed a significant difference (F₀. ₀₅, ₍₁, ₃₇₎ = 69.18, P <0.000) only in adjusted means.

Number of scale rows around caudal peduncle of M rheocharis (19-22) and M. inequalis (16-19) overlaps slightly but is significantly different. No overlap for present 17 specimens of M. rheochari : (x = 21.5, range = 20-22, SD = 0.7174) from rio Jordão (MZUSP 40279, MCP 13617, USNM 270879) and 34 specimens of M. inequalis (x = 17.6, range 16- 19, SD = 0.8170) from Morungava. Rio Grande do Sul (MAPA 115). Number of scale rows around caudal peduncle overlaps slightly in 46 specimens of M. rheocharis (x = 19.7, range = 19-22, SD = 0.7520) from rio Grande region of Santa Catarina (MZUSP 40280, USNM 306339) arrd those just listed just above of M írıequalis , but counts significantly different (t = 12.011, 9 P = 0.00).

Number of scale rows between dorsal-fin origin and anal-fm origin of M. rheocharis (17-22) and M inequalis (16-18) also overlaps somewhat but again is statistically significantly different. There is an overlap in 42 specimens of M rheocharis (x = 19.6, range = 18-22, SD = 0.8211) from the rio Jordäo (MZUSP 40279, MCP 13617, USNM 270879) and 40 specimens of M inequalis (x = 16.5, range = 16- 18, SD = 0.6405) from Morungava, Rio Grande do Sul (MAPA 115), but these counts are significantly different (t = 19.565, P = 0.00).

Lateral series scale count of M. rheocharis (41-48) and M. inequalis (36-41) overlaps somewhat but is significantly different statistically. No overlap occurs in 26 specimens of M rheocharis (x = 45.5, range = 43-48, SD = 1.4760) from rio Jordão (MZUSP 40279, MCP 13617, USNM 270879) and 38 specimens of M. inequalis (x = 38.3, range = 36-41, SD = 1.1455) from Morungava, Rio Grande do Sul (MAPA 115). Lateral series scale count overlaps slightly in 48 specimens of M rheoclıaris (x = 44.2, range = 41-48, SD = 1.7704) from rio Grande region of Santa Catarina (MZUSP 40280, USNM 306339) and those just listed of M inequalis , but these counts are significantly different (t = 17.922, P = 0.00).

Branched dorsal-fm ray count in 50 specimens of M rheocharis (x = 9.8, range = 9-12, SD = 0.6701) from rio Jordão (MZUSP 40279, USNM 270879) differed significantly from a sample of 40 specimens of M. inequalis (x = 8.125, range = 8-9, SD = 0.3349) from Morungava, Rio Grande do Sul (MAPA 115) (t = 14.951 s, P = 0.00). Same comparison for 56 specimens of M rheocharis (x = 8.9, range = 8-10, SD = 0.4615) with same sample of M. inequalis just mentioned was also significantly different (t = 9.472, P = 0.00).

Branched anal-fin ray counts overlap for both species but are statistically significantly dilferent (t = 6.336, P = 0.00) in 50 specimens of M. rheocharis (x = 25.3, range = 23-27, SD = 1.2471) from rio Jordão (MZUSP 40279, MCP 13617, USNM 270879) and 40 specimens of M. inequalis (x = 26.7, range = 24-30, SD = 0.9443) from Morungava, Rio Grande do Sul (MAPA 115). These not significantly different in a second population sample of 56 specimens of M rheocharis (x = 26.8, range = 25-29, SD = 1.3088) from rio Grande, Santa Catarina (MZUSP 40278, 40280, MCP 13616, USNM 279878. 306339) from that of M. inequalis just described.

Vertebral counts, x = 38.0, range = 37- 40, SD = 0.6558, n = 87, forM rheocharis from rio Jordão (MZUSP 40279, USNM 27879) and, x = 37.9, range = 36-39, SD = 0.7055, n = 70, for M. inequalis from Morungava, Rio Grande do Sul (MAPA 115) are not significantly different statistically.

Etymology. —The name rheocharis is from the Greek rheos (= current or stream) and charís, also Greek, (= loveliness, grace or charming) and is in reference to the streams in which this fish has been taken and to the beauty of the fish itself.

Discussion.-The phylogenetic relationships of M rheocharis are uncertain but fully mature males, as noted above, have a caudal fin-ray pump chamber. a putative synapomorphy uniting that species with M sylvicola , M. lateralis , and M microlepis . Mimagoniates inequalis and M. barberi are not included in this group because they lack a caudal pump chamber, having only a groove in its place.

Weitzman et al. (19881412) tentatively hypothesized that M. rheacharis , as new species B, is a sister species of M microlepis because these species share caudal-fin hooks (absent in other species of Mimagoniates ) distal to the pump area on principal caudal-fin rays 10 and 11. These hooks are not nearly as numerous or as large in M rheocharis as they are in M microlepis (compare Figs. 5 and 24 ). We found no other synapomorphy uniting these two species. Mimagoniates rheocharis and M. microlepis thus may still be hypothesized as one another’s closest relatives ( Fig. 1 ). However, Fig. 1 is presented primarily as a visual aid to the discussion that follows. We accept the hypothesis it presents with reservations because the relationships of M. rheocharis may be quite complex due to a possible hybrid origin or at least some introgression with other species. Furthermore a considerable amount of character conflict exists among all the species of Mimagoniates making acceptance of many of the relationships suggested in Fig. 1 quite questionable.

A number of differences, outlined in the key and diagnosis above, were found among M. rheocharis and M microlepis and M. inequalis . Although these differences indicate these species are divergent, many of these characters cannot at present be used to corroborate hypotheses of phylogenetic relationship with other species of Mimagoniates because so far they have not been identified as synapomorphies shared among other species. Certain of these characters are continuously variable between at least some species, for example between M. rheocharis and M. inequalis , possibly limiting their use as characters suitable for phylogenetic analyses, Chappill (1989) and Pimentel 8c Riggins (1987). Some of the states of these characters as shared by M. rheocharis and M inequalis are plesiomorphic with respect to their condition in other species of Mimaganiates , making their use as synapomorphies for these two species impossible. For example, outgroup comparison with species of Glandulocauda and species in other tribes of the Glandulocaudinae indicates that relatively short overall body shape and central median fin position of M rheocharis , shared with M inequalis , is plesiomorphic with respect to the more elongate body and posterior dorsal-fin position in M. microlepis , M lateralis , M. sylvicola and M. barberi (see also comments below). Relatively elongate body shapes might be considered synapomorphic for the latter four species but use of this character for corroboration of phylogenetic hypotheses in the Glandulocaudini is still in need of further study and may be difficult to use at some nodes because of continuous variability. The high number of dorsal-fin rays (8-12 in M rheocharis versus 7-9 in other species of Mimagoniates ) appears autapomorphic for that species, but, again the character is continuously variable between M. inequalis , M rheocharis and other species of Mimagoniates .

If these and other similar characters can be used as synapomorphies at some nodes, then character conflict and complex homoplasies will need to be resolved using parsimony with enough corroborated evidence to be convincing. For example, at present at least three characters or character complexes appear in conflict with one another in regard to their possible significance in suggesting phylogenetic relationships. These are relative body length, relative derivation of the caudal pump and the presence or absence of a prominent lateral mid-side dark stripe. For example, if elongate body, present in M. sylvicola , M. lateralis , M. barberi and M microlepis , were hypothesized to have evolved only once then these four species would form a monophyletic line based on this character. Then the caudal fin-ray pump chamber might have evolved twice, once in M. rheocharis and again in a clade including M lateralis , M sylvicola and M. microlepis . On the other hand, monophyly based in part on elongate body could result in a phylogeny wherein M. lateralis was the sister species to all of the other species just mentioned and M. rheocharis regained a relatively short body shape.

Again, M. lateralis and M. barberi share the possession of a dark brown or black broad lateral stripe extending from the lower jaw tip posteriorly across the body sides to the region of the caudal gland on the caudal fin. A hypothesis that considers this character synapomorphic for these two species requires that the pump chamber evolved more than once, independently in M lateralis , and in a clade consisting of M sylvicola , M rheocharis and M microlepis , a phylogenetic configuration different from those noted above for acceptance of elongate body as a synapomorphy for four species of Mimagoniates . Finally, if one hypothesizes that a fully derived caudal fin-ray pump chamber evolved only once in a clade consisting of M sylvicola . M. lateralis , M. rheocharis and M microlepis , then both the lateral stripe and elongate body would have evolved twice each and in a phylogenetic configuration difierent from any of those above. At this time the above speculations and others like them are fruitless and we mention them only to point out the current complexities regarding the phylogeny of the species of Mimagoniates. As yet we have not fully investigated the possible phylogenetic significance of many of the characters putatively available for phylogenetic analysis of the Glandulocaudini . The matter is quite complex and the use of parsimony to propose acceptable phylogenetic hypothesis and resolve the problems of character polarities will naturally still result in numerous homoplasies. We might add that the use of any “ method” other than parsimony to resolve character conflict will not eliminate all homoplasies but only subjectively delete some rather than others.

As noted above, we tentatively hypothesize that M. rheocharis and M. microlepis are sister species based on fin-ray secondary sexual characters which do not appear to be continuously variable and may be synapomorphic if the character conflicts mentioned above can be resolved in a form that favors the secondary sexual fin-ray structures. However, as one might expect, there are complications and possible alternative hypotheses of phylogenetic relationships other than those mentioned above. For various reasons some of these should be briefly discussed.

In searching for possible alternate hypotheses of phylogenetic relationship for M. rheocharis we note that this species and M inequalis look remarkably alike in general body shape, fin position, body color and the counts recorded above as separating the species are not remarkably different. Both species lack the distinct mid-lateral body stripe found in M lateralis and M barberi . Furthermore, both species have closely similar vertebral counts that are not significantly difierent, n = 39 for specimens of M inequalis from Arroio Fiuza near Passo Fiuza, Municipio de Viamão, south east of Porto Alegre, Rio Grande do Sul, (USNM 254273), x = 38.0, range = 35-39, SD = 0.6489 and n = 87 for specimens of M rheocharis from rio Jordão, (MZUSP 40279, USNM 279879), x = 38.0, range = 37-40, SD = 0.6288. See also the nearly identical results of a dilferent comparison of vertebral numbers and other counts given in the section labeled Statistical Comparisons. These vertebral counts are relatively low when compared with those of the more elongate species of Mimagoniates such as M. rheocharis but broad overlap occurs, again see section above labeled Statistical Comparisons, and we are unable to show that shared vertebral counts by M. rheocharis and M inequalis are shared derived characters. We discuss such characters only to point out the difiiculties in attempting to study the relationships of the species of Mimagoniates based on characters other than those directly associated with their secondary sexual characteristics.

Our preliminary analysis suggests that these character similarities are unlikely to be synapomorphic for M rlzeocharis and M inøqualis because these features are also found in immediate outgroups down the tree, for example species of Glandulocauda and certain of the apparently less derived glandulocaudine species belonging to tribes that are possibly sister groups to the Glandulocaudini . However, our analysis is incomplete and alternate hypotheses suggesting that M. inequalis and M. rheocharis are sister species or related in some way cannot be confidently rejected at this time.

The intermediacy of the M. rheocharis between M inequalis and M microlepis in body depth and caudal peduncle depth ( Figs. 17 , 27 ) was noted above in the Diagnosis and the Statistical Comparison section of M rheocharis . The relatively modest development of the fin-ray pump chamber and of the small and few caudal-fin spines on rays 10 and 11 in M rheocharis , rather than being considered plesiomorphic relative to the state of these structures in M microlepis , could be considered intermediate between the state in M inequalis (no spines, no pump chamber) and that in M microlepis (many well developed spines, and a highly developed fin-ray pump chamber). This suggests a possible origin of M. rheocharis by introgression between M microlepis and M inequalis . However, a hypothesis of hybrid origin for M rheocharis might be rejected because it does have autapomorphies that are absent in both M inequalis and M microlepis . Even then, it cannot be discounted that M rheocharis might have originated through introgression in a geographical range once shared by the ancestors of the current populations of M. inequalis and M. microlepis and that the autapomorphies of M. rheocharis evolved subsequent to that introgression at a time when M rheocharis became isolated from its parent populations. The current geographical ranges of these species would be congruent with such a biogeographical hypothesis but in our view this hypothesis, just as any hypothesis of species origin by introgression, unless corroborated by genetic data as well as much statistical data on morphology from pertinent natural populations, is less parsimonious than that of sister status, in this case between M rheocharis and M. microlepis . Because we lack such data, we currently reject this introgression hypothesis. If M rheocharis did evolve in part through introgression, according to our hypotheses of phylogeny among the species of Mimagoniates this introgression would have taken place between two of the more distantly related members of the genus, something we view as possible but in need of study by breeding experiments and intensive statistical analyses of the characters of the three species from population samples in the areas of their contingency and sympatry.

It might be proposed that specimens of M rheocharis are simply hybrid individuals between M irıequalis and M. microlepis but several factors argue against this. First, only one of the putative parent species occurs in the present range of M rheocharis , that is M microlepis which is sympatric and syntopic with M rheocharis . Mimagoııiates inequalis , as explained above. is so far as known allopatric with M rheocharis , preventing it from being a parent of current specimens of M. rheocharis . Furthermore. M rheocharis has autapomorphic characters not present in either putative parent.

Another hypothesis of phylogenetic relationship, one that is less parsimonious than sister species status for M. rheocharis and M microlepis , could be put forward and we mention this one only because of its implications regarding the dangerous practice of assuming biogeographical significance for geographically adjacent similar looking species, done by some biogeographers. One could hypothesize that M. rheocharis and M inequalis are sister species. This hypothesis is not out of the realm of possibility even though current character polarity hypotheses for secondary sexual characters irrdicate that other features shared by these latter two species are plesiomorphic at this level of relationship. Nevertheless, if such a hypothesis were true the current distributions of these two species would make good historical biogeographical sense. Mimagoniates inequalis is found in the freshwater biogeographical South Coastal Subregion of Menezes (1988: 300) and M rheocharis is found in the adjacent southern part of his Central Coastal Subregion. A simple vicariance or dispersion event historically separating two parts of an ancestral population for the two species could have been responsible for initiating their evolution. However, the phylogenetic evidence, in all its current uncertain status, does not support such a hypothesis of phylogeny and therefore does not support this historical biogeographical pattern. This example provides one reason why we refuse to use current distributional patterns of species as either evidence of relationships among organisms or as direct indicators of historical biogeographic events. Most distribution patterns may correlate with a variety of phylogenetic and biogeographic scenarios and cannot be used as primary evidence for any given one.

Another problem with some of our data remains to be discussed. Given the current limitations of our population samples and the small adult lengths of most of our specimens of M iııequalis , it is difficult to place too much confidence in certain aspects of the above phylogenetic discussions based on some secondary sexual characteristics. This is true because at least one, caudal pump morphology, of the phylogenetically important characters currently available for M. rheocharis and its possible relatives such as M. inequalis , is based on a character that may be juvenile, that is not representing a character state attained in fully mature adults of that species. Such juvenile character states may then be compared incorrectly with homologous but fully adult characters in another species, leading to invalid conclusions regarding the relative apomorphy of these characters. This kind of problem affects both of the very separate tasks of studying phylogenetic relationships and of distinguishing species from one another, witness our caution about couplet four in our key above and our mistake in Weitzman et al. (1988) regarding the relative apomorphy of the caudal gland in M. sylvicola (see the Discussion above of that species). Regarding M. inequalis in the current treatment, perhaps our collections contain no fully adult males of this species and at the completion of sexual maturity, males have a caudal fin-ray pump chamber similar to that found in males of M. rheoclıaris . If this were true the phylogenetic relationships we have presented based on caudal morphology would be drastically altered. The evidence we have bearing on this problem is as follows. A male specimen of each of the two species of nearly equal length (one, M. inequalis , SL 38.8 mm, MCP 9892, from rio Cai drainage, Rio Grande do Sul, and the other, M rheocharis, SL 39.5 mm, MCP 10806. from rio Três Furquilhas, Rio Grande do Sul), indicate that M rheocharis has a well-developed caudal-fin ray pump chamber as well as tiny hooks on principal caudal fin-rays 10 and ll and that M inequalis only has a groove and no hooks. Also, two old aquarium specimens of male M ineqııalis (USNM 94310, originally imported from Porto Alegre and grown to “ old age " in an aquarium by Herman Meinken) have standard lengths of 38.2 and 41.0 mm. The latter specimen is the largest we have of this species. These specimens show no evidence of a caudal pump chamber or hooks on fin-rays 10 and ll but do have a groove. Furthermore, three smaller (SL 28.5-31.3 mm) sexually mature males of M rheocharís (MCP 13613, from arroio Ägua Parada, a tributary of the rio Maquiné from near Maquiné, Rio Grande do Sul) have well-developed pump Chambers as well as tiny hooks present on caudal-fin rays 10 and ll. These facts reinforce our hypothesis that M. rheocharis is a derived species distinct from the more plesiomorphic M. inequalis .

The specimens of M rheocharis , MCP 13613, three males and four females, from arroio Agua Parada plus a single large specimen (SL 41.8 mm, MZUSP 40281, Fig. 20 , from a nearby locality, arroio das Pedras near Osório), are interesting in light of the introgression hypotheses discussed above. The meristic characters that help distinguish M. inequalis and M. rheocharis are all within the overlapping range of the two species (9 branched dorsal-fin rays, 16-17 scale rows between the dorsal-fin and anal-fin origins, 18-19 scale rows around the caudal peduncle and 42-43 scales in a lateral series). Although these counts fall well within the range of those of other population samples of M rheocharis to the north, they do fall closer to those counts of M. inequalis than any of our other population samples of M rheocharis . Compare the counts just given with those presented in the comparative section of the Diagnosis of M. rheocharis above. We also note that these localities for M. rheocharis near Osório lie adjacent to the rio Gravataí drainage just to the west which is occupied by M inequalis . We are unsure of the real significance of these counts but they could imply some gene flow between M. rheocharis and M. inequalis in this region at some time in the past. We are unaware of any current sympatry of these two species in this region but the area is not well sampled.

In summary. the phylogenetic relationships of M. rheocharis are uncertain. It could simply be a sister species of M. microlepis with a somewhat more plesiomorphic state of the caudal pheromone origin, a somewhat derived sister species of M inequalis , a species derived by introgression between M inequalis and M. microlepis with subsequent evolution of some autapomorphic characters, or, for example, a sister species of M. microlepis which may be introgressing with M inequalis . Other hypotheses of phylogenetic relationship could be suggested but these examples are sufficient to indicate the possible complexities of the phylogeny of M rheocharis . Because the shared characters of M. rheocharis and M. inequalis have been hypothesized to be pleisomorphies in the context of the entire phylogeny of the Glandulocaudini , we, at least for now, accept the evidence that the male caudal secondary sexual characters represent synapomorphies and tentatively propose sister species status for M rheocharis and M. microlepis . The other hypotheses of possible phylogenetic relationship await considerable additional evidence for possible corroboration.

Ecology. -We do not have extensive notes on the ecology of this species but the following information is taken from notes by Carlos Lucena from two localities (one the type locality) and by Ricardo M. C. Castro from two localities.

Specimens collected by C. A. S. de Lucena and party from rio Faxinalzinho at Mâe dos Homens, rio Mampituba drainage, near Praia Grande, the type locality, MZUSP 40278, MCP 3616, USNM 279878, were removed from a clear-water stream 2-6 m wide, 0.1-1.0 nl deep, with a moderate current. There was a moderate amount of (unspecified) marginal vegetation present. The substrate consisted of rubble, rocks, stones, and gravel. Mimagoniates microlepis and a species of each of Rhamdia, Rlıamdella, Heptapturus , Ancistrus , Rhineloricaria, Pseudotocirıclus, Pareiorhyna, Corymbopharıes , Trichomycterus, Characidum, an unspecified species of tetragonopterine characid, two species of Astyanax , and a species of Jenynsia were present.

The second locality collected by Carlos Lucena and party was rio Jordão near Jordão Alto, MCP 13617, MZUSP 40279, and USNM 270879. This was the upper region of the rio Jordão where the stream was 1- 5 m wide and 0.1 -0.7 m deep. The current was slow, there was a small amount of (unspecified) marginal vegetation, and the substrate consisted of rubble, rocks, stones, gravel, sand and mud. The fishes that were collected at the same locality were species of Rhamdia , Heptapterus , Ancistrus , two loricariids unidentified to genus, Rhineloricaria , Pseudotocinclus , two species of Trichomycterus, Characidium . Hoplias , Astyanax , an unidentified tetragonopterine characid, Phallaceros, Jenynsia , Geophagus brasiliensis and Cichlasoma facetum .

The first locality collected by Ricardo Castro is from near Osório, Rio Grande do Sul, MZUSP 40281. The specimen was collected in a small clear-water stream about 1.5 m wide and 20 cm deep. surrounded by shrubs and small trees. The water was clear, with very little suspended matter. Algae covered rocks were common on the substrate. Other fishes collected in the area were species of Astyanax, Oligosarcus , Rhamdia , and a small unidentified catfish.

Specimens of M rheoclıaris from a tributary of the rio Grande, Santa Catarina, MZUSP 40280 and USNM 306339, were collected from a slow moving clear water stream about 1.5 m wide and 30 cm deep. Not far away the stream flowed into the larger river. Rocks, sand, soil and twigs fallen from nearby trees were the main substrate components. The rocks and twigs were densely covered with algae. The stream was bordered by rocks, mud, grass and small trees in many places.

Figures

Fig. 17. Mimagoniates rheacharis, M. microlepis and M. ínequalis, body depth as a function of SL by species and sex. Plot axes are logarthmic. Logarithmic regression equation for 31 males of M rheocharis: (Y = -1.846 + 1.199X), 25 females or this species: (Y = 2.115 + 1.27lx); for 39 males ofM. microlepís: (Y = 2.130 + 1.235X) and for 31 females of the later species: (Y = -2.705 + 1.399 X); and for 26 males of M inequalis: (Y = -2.230 + 1.3499X), 14 females of this species: (Y = -1.812 + l.2l87X). For explanation see Diagnosis, Sexual dimorphism, Discussion and Description under M. rheocharís.

Fig. 18. Mimagoniates rheocharis, new species, holotype, MZUSP 40278, male, SL 47.3 mm; Brazil, Santa Catarina, rio Faxinalzinho at Mãe dos Homens, near Praia Grande.

Fig. 19. Mimagoniates rheoclıaris, new species, paratype, USNM 306339, female, SL 34.3 mrn: Brazil, tributary to rio Grande, Santa Catarina.

Fig. 20. Mimagoniates rheacharix, new species, paratype, MZUSP 40281, male, SL 41.8 mm; Brazil, Río Grande do Sul, arroio das Pedras, near Osório.

Fig. 21. Mimagoniates rheocharis. new species, paratype, anterior 13 anal-fin rays of an adult male, lateral view, anterior is at lefi. USNM 279879, SL 45.4 mm; Brazil, Santa Catarina, rio Jordão at Jordão Alto. For explanation see Description under M. rheocharis.

Fig. 22. Mimagoniates rheocharis, new species, paratype, pelvic fin of adult male, ventral view, left side, anterior at left, medial fin ray at bottom. USNM 279879, SL 45.4 mm; same locality data as specimen in Fig. 2 l.

Fig. 23. Mimagoniates rheocharis, new species, paratype, unstained central basal region of caudal fin and adjacent caudal peduncle of adult male, lateral view, left side, anterior at left. USNM 279879, SL 49.0 mm; same locality data as specimen in Fig. 21. Arrow at lefi indicates anterior intake region of pump chamber. Middle arrow indicates region of lateral slit with ray half of principal ray 11 just dorsal to arrow tip. Arrow at right points to opening of posterior exhaust vent of pump chamber. Note that in this species pump chamber not formed into an obvious cylindrical tube as that of M. microlepir, Fig. 3. Glandular tissue surrounding chamber openings in this photograph obscure details of pump.

Fig. 24. Mimagoniates rheocharis, new species, paratype, osteology ofcentral basal region ofcaudal Skeleton, principal fin rays 6-15, of an adult male, lateral views, left side, anterior at left. USNM 279879, SL 45.4 mm, same locality data as specimen in Fig. 21. Principal ray 12 indicated by arrow. (A) Illustrates caudal pump region. (B) illustrates relationship of modified dorsal caudal-fin lobe squamation to caudal pump.

Fig. 25. Mimaganíates rheocharis, new species, paratype, jaws and dentition of adult male, lateral view, right side, anterior at right. USNM 279879, SL 45.4 mm, same locality data as i.n Fig. 21. For explanation see text Description under M rheocharis.

Fig. 26. Mimaganiates rheocharis, new species, paratypes, body depth as a function of SL by sex. Plot axes are logarithmic. Logarithmic regression equation for 27 males of M. r/ıeocharís: (Y = -2.190 + 1.300X) and for 23 females of this species: (Y - -1.831 + 1.183 X). For explanation see Sexual dimorphism under M rlıeocharis.

Fig. 27. Mimaguniates rheacharis, M. microlepis and M. inequalis caudal peduncle depth as a function of SL by species and sex. Plot axes are logarithmic. Logarithmic regression equation for 31 males of M. rheocharis: (Y = -4.222 + 1.588 X), for 26 females of the same species: (Y = —3.311 + 1.374X); for 40 males ofM microlepís: (Y = —4.151 + 1.592X) and for 30 females of the later species: (Y = -3.239 + 1.374X); and for 26 males ofM inøqualis: (Y = —3.021 + 1.3375 X), 14 females ofthis species; (Y = -2.625 + 1.2ll 6 X). For explanation see Diagnosis, Sexual dimorphism, Discussion and Description under M. rheochuris.

Fig. 28. Mimagoniater rheocharir, new species, paratypes caudal peduncle depth as a function of SL by sex. Plot axes are logarithmic. Logarithmic regression equation for 27 males of M. rheocharis: (Y = -3.490 + 1.447X) and for 23 females ofM. rheocharis: (Y = -2.606 + 1.167 X). For explanation see Sexual dimorphism under M rheocharís.

Fig. 2. Mimagoniaıes barberi. unstained central basal region of caudal fin adjacent to caudal peduncle of a preserved adult male, lateral view. left side, anterior at left. SL 32.7 mm. UMMZ 205420, Paraguay, rio Aguarymi, tributary to rio Paraguay, San Pedro. Illustrates external features of a caudal gland without complex pump chamber. Darkly pigmented glandular groove between principal rays ll and 12 indicated by arrow. Modified dorsal caudal-fin lobe scales, although present, are transparent and not visible.

Fig. 5. Mimagoniates microlepis, osteology of central and basal region of caudal skeleton of adult male, lateral view, left side, anterior at left. SL 45.3 mm, USNM 236089, Brazil, Parana, rio Nhundiaquara at Morretes. Relationship of modified dorsal lobe caudal-fin squamation to caudal gland shown in inset at left. Distal regions ofventral and posterior scales and their epidermis form a flap partly covering intake openings of pump chamber.

Fig. 3. Mimagoniates barberi, osteology of central basal region of caudal-fin skeleton, principal fin-rays 6- 15, of adult male, lateral views, left side, anterior at left. SL 35.6 mm, UMMZ 205420, same locality as Fig. 2. Principal ray 12 indicated by arrow. (A) Illustrates glandular groove between rays ll and 12. (B) Illustrates relationship ofmodified dorsal caudal-fin lobe squamation to glandular groove. Note that modified scales together with their epidermis form a movable flap just lateral to surface of rays and glandular groove. Free border of this flap occurs along ventral margin of ventral row of scales and at posterior edge of posterior scale of ventral scale row.

Fig. 1. Diagram of tentative phylogenetic relationships of the species of the Glandulocaudini based on an analysis of synapomorphies found in the caudal skeleton of males. Filled in squares = apomorphies and empty squares - plesiomorphies at the levels indicated.

Tables

Table 2. Morphometrics of Mímaguııiates rheocharis, new species. Standard length is expressed in mm; measurements through head length are percentages of standard length; the last four entries are percentages of head length. Specimens are from near Rio Grande, Santa Catarina, MZUSP 40278, 40280, MCP 13616, and USNM 279878, 306339.

	Hololylle	n	Range	î:
Standard length	47.3	57	13.5-47.8	29.5
Depth at dorsal-fin origin	—	57	22.2-35.2	30.5
Males	33.8	31	26.9-35.2	31.4
Females	—	26	22.2-32.9	29.4
Snout to dorsal-fin origin	57.9	57	54.6-60.7	58.0
Snout to pectoral-fin origin	26.6	57	24.2-29.6	26.3
Snout to pelvic-fin origin	43.4	57	40.4-47.6	45.0
Snout to anal-fin origin	58.8	57	52.3-58.8	56.9
Caudal peduncle depth	—	57	8.9-16.3	13.4
Males	16.0	31	11.5-16.3	14.2
Females	—	26	8.9- 14.2	12.4
Caudal peduncle length	1 1.4	57	7.1-13.3	9.9
Pectoral-fin length	21.6	57	18.5-24.9	22.7
Pelvic-fin length	12.3	57	11.1-15.1	13.4
Dorsal-fin base length	17.3	57	12.7-19.1	15.8
Dorsal-fin height	26.0	57	20.7-28.0	23.7
Anal-fin base length	34.5	57	33.7-38.7	36.0
Anal-fin lobe length	19.0	57	15.6-22.6	19.9
Eye to dorsal-fin origin	45.9	57	41.5-48.3	45.2
Dorsal-fin origin to caudal-fin base	46.5	S 7	39.3-49.2	45.4
Bony head length	25.6	57	24.2-28.1	25.5
Horizontal eye diameter	33.1	S 6	33.1-48.1	39.5
Snout length	24.8	56	20.0-26.2	22.5
Least interorbital width	33.9	S 6	28.9-40.4	36.0
Upper jaw length	43.8	56	42.9-50.0	45.7