Poecilia (Acanthophacelus) obscura, Schories, Susanne, Meyer, Manfred K. & Schartl, Manfred, 2009

Schories, Susanne, Meyer, Manfred K. & Schartl, Manfred, 2009, Description of Poecilia (Acanthophacelus) obscura n. sp., (Teleostei: Poeciliidae), a new guppy species from western Trinidad, with remarks on P. w i n g e i and the status of the “ Endler’s guppy ”, Zootaxa 2266, pp. 35-50 : 39-48

Poecilia (Acanthophacelus) obscura , new species

( Figs. 2 a–c, 3, 4, 6–8, table 2)

Holotype. Male ( MTD F31503), 15.1 mm SL, for other measures see table 2; Oropuche River (N 10° 43.052’ W 61° 8.871’) below the bridge where the road from Cumaca to Cumaca Caves crosses the river approx 2.5 km past Cumaca, Trinidad; February 13th 2008, S. Schories, M. Schartl, P. Fischer leg.

Paratypes. 3 males, 5 females ( MTD F31504 – 31511); for measures see table 2; same collection as holotype. 4 males, 2 females ( SMF 31068); for measures see table 2; Rio Seco, below Río Seco waterfall, at junction with Salybia River, Trinidad, February 17th 2008, S. Schories, M. Schartl, P. Fischer leg.

Diagnosis. Poecilia obscura ( Fig. 2 a–c, 3, table 2) is a small poeciliid species from the subgenus Acanthophacelus (SL in males usually not exceeding 20 mm and in females 30 mm), which is distinguished from all other species of the subgenus by the following characters: comb-like spines of gonopodium ray 3 not numerous (11 to 15), basal massive, vs. numerous (14 to 18), basal slender in P. reticulata and P. wingei (Fig. 3). Being genetically divergent from the other species of the subgenus, the definition of the new species is mainly based on evidence from DNA sequence.

P. obscura is also recognized by morphometrics: caudal peduncle short in females (SL/CPD = 4.8 to 6.15), vs. long (SL/CPD = 6.05 to 7.4) in P. reticulata and P. w i n g e i (SL/CPD = 6.5 to 7.3) ( Fig. 4 ); usually 6 dorsal fin rays, vs. usually 7 in P. reticulata and P. wingei .

Description. Body deep in females, head long and not sharply pointed, 25.5–26.5% of SL. Longitudinal scale series 24 to 25 (rarely 24); predorsal scale series 12 to 13; scale series around caudal peduncle 14. Number of vertebrae 25 to 27. Gill rakers on first arch 11 to 12.

Teeth of upper and lower jaws unicuspid and recurved; those of outer row enlarged and spatula-shaped, numerous and widely spaced, the much smaller posterior teeth have a single sharp spine. Upper pharyngeal bones balloon shaped. Teeth of the medial region somewhat enlarged, each side with a series of 8 rows, teeth small and conical. Lower pharyngeal bone (ceratobranchial 5) heart-shaped and with very numerous pointed and slender unicuspid teeth, 17 to 19 on posterior rows, 5 to 6 on middle rows. Teeth of medial region somewhat enlarged. Arms of pharyngeal very short and split at the ends. Ceratobranchial 4 without teeth, hypobranchial 4 present.

Supraorbital canal system well developed, section 1 to 2a and 2b to 4a, usually represented as grooves, 6b to 7 as pits; preopercular canal with 6 pores; preorbital canal with 4 pores (sometimes represented as groove).

Gonopodium (Fig. 3) short and compact, 2.9 to 3.1 times in SL; ray 3 broadly expanded, curved distally and with widely spaced, 11 to 13 comb-like spines not terminating in a hook; ray 4a curved distally, with a small serra, well developed broad and long segments; ray 4p distally somewhat curved and with massive developed serrae, proximal serrae with 3 to 5 thorns, rays 5a and 5p straight, distal with a large recurved boomerang-shaped bony hook. Rays 6 and 7 long and somewhat thickened distally and curved dorsally.

Gonopodial suspensorium with two well developed gonapophyses, gonapophyses I and II long and slender, each with an uncinus. Ligastyle absent or not well developed. Gonactinost 1 with a small inferior wing-like appendage, gonactinostal complex 2 to 4 in front with a superior small lateral wing, gonactinosts 5 to 9 without bony plates or outgrowths.

Small dorsal fin with 6 to 7 rays (first ray simple, all others branched), origin of dorsal fin posterior to the insertion of anal fin; caudal fin with 22 to 23 rays (10–13 rays branched); anal fin 9 rays (6 rays branched); pectoral fin with 14 rays (9 rays branched); ventral fin with 6 rays (first and last ray simple, all others branched), in females not reaching to the anal fin base and in males reaching widely over the base of the gonopodium.

FIGURE 3. Gonopodium of P. o b s c u r a Males and females with sex specific coloration ( Fig. 2 a,–c). Body color of adult females grayish brown to yellow, fins hyaline; body sides of adult males with red, blue, orange and yellow bright pigment spots, some reflecting iridescent, usually with 1 to 3 rounded black spots, sometimes with a series of irregularly thin and short or long brown or light black horizontal lines or with very short brown vertical lines sometimes crossing the horizontal bars; the caudal fin base often shows a lower or upper black spot surrounded by small dark and short dashes and yellow pigment, dorsal or ventral caudal fin rays sometimes pigmented and rarely elongated over the caudal margin of the fin, forming a short “sword”; dorsal fin often whitish, dark or polychromatic colored, sometimes flag-like elongated, all other fins hyaline. Male body coloration extremely polymorphic: in natural habitats no two males being alike. Body coloration and caudal appendix phenotype predominantly heritable characters, male offspring of a single male in laboratory crosses being very similar to their father and to each other.

# TL SL HL SNL BD IOW GL ED APL CPL CPD PL VL PDL HT(M) MTD F 31503 21.3 15.1 4.2 1.2 4.1 1.9 4.4 1.3 4.9 2.9 3.5 9.9 PT(F) MTD F 31504 33.4 25.9 6.0 1.4 7.3 3.4 2.1 9.6 8.9 4.4 5.6 3.3 16.6 PT(F) MTD F 31505 35.8 28.3 6.2 1.5 7.7 3.6 2.2 9.9 9.3 4.7 5.6 3.4 17.1 PT(F) MTD F 31506 32.6 25.5 5.7 1.4 7.2 3.3 2.1 8.3 8.0 4.3 5.3 3.2 16.3 PT(F) MTD F 31507 28.9 21.7 5.3 1.3 5.8 2.8 1.8 3.8

PT(M) MTD F 31508 24.4 17.8 4.6 1.3 4.8 2.3 4.6 1.5 3.7 6.4 3.3 3.7 10.6 PT(F) MTD F 31509 21.1 16.2 4.3 1.3 4.2 2.1 1.4 3.2 10.2 PT(M) MTD F 31510 21.4 15.3 4.2 1.3 4.2 2.0 4.5 1.4 2.7 4.2 3.6 9.8 PT(M) MTD F 31511 19.2 13.9 4.0 1.1 3.9 1.7 4.2 1.2 4.6 2.7 4.0 3.4 9.6 PT(M) SMF 31068 24.0 17.7 4.5 1.3 4.9 2.3 4.7 1.5 3.8 6.3 3.3 4.9 3.7 10.7 PT(F) SMF 31068 32.7 25.4 6.1 1.4 7.0 3.4 2.1 9.7 9.4 4.2 5.4 3.3 16.4 PT(F) SMF 31068 33.5 26.1 7.2 3.5 2.1 9.5 9.5 4.5 5.5 3.3 16.5 PT(M) SMF 31068 25.6 18.2 4.7 1.3 5.0 2.4 4.9 1.6 3.6 3.9 10.9 PT(M) SMF 31068 24.4 17.9 4.5 1.2 2.3 4.7 1.5 3.8 6.2 3.5 3.8 10.5 PT(M) SMF 31068 20.4 14.7 1.1 4.0 1.8 4.3 1.3 2.8 9.4

all values are in mm, HT holotype, PT paratype, M male, F female.

Etymology. The name of the new species is derived from the latin word obscurus (hidden) because of its status as a cryptic species.

For the common name we propose to call P. obscura the Oropuche guppy due to its main area of distribution. For P. reticulata, Poeser et al (Poeser et al., 2005) suggested the name Common or Orinoco guppy. For P. w i n g e i they proposed the name Campoma guppy; reflecting the narrow range on the Paria peninsula where they found the species. After it is now clear on the basis of our molecular analysis that the species also occurs in the Cumaná region and that it represents Endler’s guppy, it makes sense to return to the priority common name of Cumaná or Endler’s guppy ( Alexander and Breden, 2004).

Habitat notes. P. obscura was collected from fast flowing small rivers and ditches with generally clear waters. The species also occurs in downstream habitats. Thus it inhabits both so-called low and high predation sites described in the guppy literature (see Magurran, 2005). The locus typicus ( Fig. 5) is a site in the headwaters of the Oropuche River. The river is here about 5 meters wide and fast flowing with a maximum depth of 1 meter. The riverbed is gravel and sand with some larger rocks in the middle. Submerse plants are absent, but the vegetation on the river banks focally reaches into the water and provides shaded sections throughout most of the day. On the 13.2.2008 (1 p. m., air temperature 25.5°C) the following parameters were recorded: clear water, water temperature 22.5°C, hardness 5–10, conductivity 51 mS, pH 7, nitrate 0, nitrite 0. Accompanying fishes were Rivulus hartii , Astyanax spec ., Ancistrus spec . and small cichlids. The inclusion of cichlids and characins defines the habitat as high predation site.

Comparison, relationships and discussion. On the basis of synapomorphies, P. obscura is unequivocally attached to the subgenus Acanthophacelus . By morphological criteria P. obscura is most closely related to P. reticulata and P. wingei . There are several synapomorphies that unite the three species, namely: large recurved hook present on tip of gonopodium ray 5 and serrae of ray 4p and 4a well developed, long and pointed comb-like spines of gonopodium ray 3 starting very close to the tip. Male pigmentation colorful and extremely polymorphic, more brilliant, metallic and less variable in P. w i n g e i. We noted no difference in the morphology of the genital pore of the females from the three species ( Fig. 6 ).

However, P. obscura is recognized as a separate species, because it does not share the following characters with P. reticulata and P. w i n g e i: fewer subdistal comb-like spines of gonopodium ray 3 (11–15, vs. 14–18), higher and shorter caudal peduncle in females, apparent as lower SL/CPD values ( Fig. 4 ). In P. obscura SL/ CPD values of males are in the same range as females while there is a sexual dimorphism for this value in P. reticulata . In this species the SL/CPD ratios of males are lower than for females and in the same range as males of P. obscura . Whether also P. w i n g e i is polymorphic for this character could not be conclusively determined from the available specimen and requires inspection of the type material in the future.

The female of the paratypes of “ Girardinus guppii ” (ZMB 6081) that were sent by Dr. Günther to his colleague at the Museum in Berlin was also measured. These fish are designated as” origin Trinidad ”, however, no further information is available on the exact collection place or even river system. The SL/CPD value of this fish is close to the average for P. reticulata . It has to be considered, however, that with respect to morphometrics substantial shape differences have been noted between different populations ( Alexander et al., 2006; Hendry et al., 2006). Only when more populations of P. o b s c u r a can be analyzed in the future it will become clear how robust these characters are and how suitable they will be for determining the taxonomic status of the species.

A molecular phylogenetic analysis was performed using mitochondrial D-loop and cytochrome b sequences. The molecular data were analyzed with maximum likelihood, parsimony and neighbor-joining methods which yielded almost identical phylogenetic results. The topology of the resulting trees was always the same.

The phylogram rooted on Micropoecilia picta has three well-separated branches ( Figs. 7 , 8 ), which are supported by high bootstrap values. One branch is built by the sequences of P. wingei from the locus typicus and several collections from the city of Cumana. A second branch consists of the P. obscura sequences. The third branch represents P. reticulata . Here all other guppy sequences cluster that span the entire mainland range of the species from Columbia to Brazil and the islands of Tobago and Isla Margarita. Importantly the West Trinidadian guppies from the Caroni drainage are also firmly nested in this branch.

We also obtained sequences from a specimen of P. reticulata (ZMB 6299) of the collection of Julius Gollmer from which the type material for the first description was taken. This fish was collected and preserved on the 14.5. 1856 in the Rio Guayre near the city of Caracas ( Venezuela). This sequence clusters with all present-day P. reticulata including a fish taken in 2008 from a tributary to the Guayre (RG) near Caracas ( Fig 8 ).

It is a well-known fact that guppies evolve extremely rapidly in response to natural selection ( Magurran, 1998; Reznick et al., 1997). This phenomenon could be partially explained by a very high natural mutation rate, which in turn could compromise the significance of the molecular analysis. We therefore included in our dataset five different ornamental guppy strains. These fish independently have been under the most extreme selection for several polygenic traits under enforced short generation times by guppy breeders in Russia, Taiwan, Thailand and Singapore. On the molecular level they were, however, indistinguishable from wildcaught guppies ( Fig. 8 ).

We included also a laboratory strain of P. reticulata that is kept as closely inbred line and displays a very homogeneous male pigmentation phenotype. This strain, designated maculatus-zebrinus (ma-ze), was established by Ojvind Winge in the twenties of the last century. Its mitochondrial DNA sequence is nested within all the other P. reticulata .

The new species is described mainly on the basis of molecular data. Although the differentiation of the three branches of guppies is unequivocal and indicates separated (mitochondrial) gene pools it may be questioned whether the genetic distance is in the range that usually is found between species that are more classically defined by morphological criteria. To evaluate this, the genetic distances and branch lengths between the three guppy species were compared to the values obtained in studies of other poeciliid fish groups where also mitochondrial control region and the cytochrome b gene were used for phylogeny construction. The genetic distance separating the three guppy taxa is 0,0 29. This is well in the range of values obtained for morphologically very different species in the genus Xiphophorus (0.029 between Southern platyfish species, X. maculatus , X. milleri , and Southern swordtails, e.g X. hellerii , X. signum ( Meyer and Schartl, 2002) ; 0,0 18 between the Southern swordtail species, e.g. X. hellerii , X. signum , X. alvarezi ( Meyer and Schartl, 2003) ; 0,0 25 between Priapella olmecae , chamulae , compressa and intermedia ( Schartl et al., 2006); 0,0 38 between the shortfin mollies of the genus Poecilia ( Meyer et al., 2004)) .

Because the three species of the subgenus Acanthophacelus are not readily or reliably distinguishable on a morphological basis but can be separated using non-morphological data, such as from DNA sequence analysis, they can be regarded as another example of a cryptic species complex.

Although the validity of estimating divergence times from DNA sequence variations has been debated (e.g. Howell et al., 2008) it can give at least a reasonable estimate, in particular when calibrations are available from related groups of organisms as is the case for the mitochondrial control region and cytochrome b genes of actinopterygian fishes (for a recent compilation see Burridge et al., 2008). Using the fastest and slowest rates of all known calibrated molecular clocks in teleosts the separation of the three guppy species should have occurred between 2.5 and 5 million years ago based on the cytochrome b sequences (0.0076 – 0.0036 changes/site/Myr) and between 0.4 and 4.2 million years based on the sequences of the control region (0.044 – 0.004).

P. o b s c u r a is endemic to the island of Trinidad. A genetic divergence of the guppies from the Oropuche drainage has been noted earlier using mitochondrial DNA sequences or allozyme data, ( Alexander and Breden, 2004; Carvalho et al., 1991; Russell and Magurran, 2006); ( Alexander et al., 2006). In all phylogentic trees published so far by these authors the so-called “Oropuche guppies” formed a separate branch. But this was never discussed on basis of recognizing these fish as a separate species. Moreover the differentiation between the fish from the Caroni and the Oropuche drainages was interpreted as the two subgroups being on their way to establish different species, but not having completed the whole process. Our analyses presents a different view, namely of two species that have been separated since a long time. The Caroni drainage, which drains west into the gulf of Paria is separated by a watershed from the Oropuche drainage although the minimum distance between the two river systems during the rainy season can be as close as 70m ( Magurran, 2005). The Oropuche and some other rivers with P. obscura flow east and drain into the Atlantic. Notably, there is a third drainage, namely those rivers that drain north into the Caribbean Sea. Interestingly these fish, although clearly representing the species P. reticulata somehow cluster in the phylogenetic trees with P. reticulata from the northward-located island of Tobago ( Fig. 8 ), indicating a possible colonization from Tobago towards Northern Trinidad or vice-versa.

The hypothesis that the Caroni (including the Northern rivers) and Oropuche drainages have an independent freshwater fauna from different ancestral lineages is supported by studies on reptiles ( Boos, 1984), the cyprinodont fish Rivulus hartii ( Jowers et al., 2008) and paleogeographic evidence. During the glaciation periods of the Pleistocene sea levels were up to 130 m lower and Trinidad was connected to the mainland ( Kenny, 1988). The Orinoco, the main distribution area of P. reticulata , at that time reached further northward and was probably connected to the Caroni drainage, discharging to the Gulf of Paria. The Oropuche drainage had most likely confluence with a separate more easterly river system and was colonized from there. Thus P. o b s c u r a and P. reticulata arrived from different regions to Trinidad and did not cross the watershed that still separates them in the Northern mountain range. We hypothesize that P. wingei was the most western species but recently was circled by northwest colonizing populations of P. reticulata .

Poeser et al. (2005) argued that the Eastern Trinidad guppies, now P. obscura , originally were the same species as P. w i n g e i and have been “despeciated” through hybridization and genetic introgression by the Common (or Orinoco) guppy, P. reticulata . Our molecular analysis does not support this interpretation, as there is absolutely no indication of a closer relationship of P. obscura with P. wingei .

Given the strong variation in male pigmentation the description of P. wingei on the basis of coloration was not conclusive. But our molecular data now define the species unequivocally and show that it is a separated taxon from P. reticulata and P. obscura . Clearly, all fish earlier classified and recognized as Endler’s guppies from localities in and around Cumana are P. wingei .

The molecular barcode from mitochondrial sequences will help to define the actual species ranges and hybrid zones. Evidently fish from collecting station 6 of Poeser et al. (2005) were designated by these authors as P. reticulata on the basis of their male coloration phenotype. We revisited this place (El Cordon, EC). The molecular analysis revealed clearly that these fish have the P. wingei mitochondrial sequence ( Fig. 8 ). No further data exist to clarify whether this discrepancy can be explained by interspecific hybridization. We also noted as reported by Poeser et al. (2005) that in the river below the El Cordon waterfall several males had an unusual long gonopodium. However, closer inspection revealed that all of these were immature males and that the gonopodium, once fully differentiated is not longer than of any other guppy. Only in the transition period of the anal fin to the intromittent male genitalia a prolongation of the fin happens. This can be seen - although in a less extreme degree - in any other guppy strain. It is unclear whether this is an environmental effect, because such an extreme lengthening of the transforming anal fin was not observed anymore in the subsequent captivity reared offspring.

Male coloration is much more uniform in P. wingei from Cumana and is much alike the pigmentation of the fish from Campoma and all earlier known Endler’s guppies ( Fig. 2 g,h). We suspect that fish (see Fig. 2 f) collected by Poeser et al. (2005) including those from the El Cordon waterfall, which have a “similar color pattern as the Orinoco variety” (= P. reticulata ), but have additionally some metallic pigment spots like P. wingei from the Locus typicus and the Cumana populations (CC, LP) may be hybrids. Future studies that include nuclear genetic markers will be necessary to clarify this.

All other fish that based on molecular data could be unequivocally assigned to the P. wingei clade were of a typical coloration distinct (see also Fig. 2 g,h) from P. o b s c u r a and P. reticulata , which we describe here as: longitudinal broad red band extending maximally from the operculum to the caudal fin base, along or above the lateral midline in the trunk and along or below the lateral midline in the peduncle, often interrupted by a vertical black bar originating around the anal fin base and extending to the dorsal fin base or anterior thereof. The red band often dissolved into oval or rarely round red spots, especially in the peduncle. A black stripe above the red band extending maximally to the eyes and the caudal fin base, sometimes missing in the trunk area. Additional dark black coloration of the ventral margin of the peduncle in many males. Caudal fin often with ventral and dorsal sword-like coloration, red, yellow or white, often with a black margin, frequently upper and/or lower rays of caudal fin prolonged to form colored swords or double-swords. Some males with large black shoulder spot with fuzzy margin, similar to the vertical bar. Perfectly round dark black body spots with sharp borders, which are typical for P. reticulata and P. obscura males, very rare. Large roundish or oval bright metallic blotches of, green, or more rarely yellow or light blue iridescent color following the basic longitudinal pattern interrupting the red band in the peduncle. White markings rare on the body sides. Rarely small black irregular spots on the belly. Dorsal fin hyaline, yellow and black coloration, sometimes anterior and dorsal margin black. This typical coloration has been documented extensively by Alexander & Breden (2004) for the populations from Central Cumana and we found this also for the Laguna de los Patos (West Cumana) and Campoma populations.

P. o b s c u r a was most probably not recognized so far as a separate species, because these fish are not different from P. reticulata in the most obvious character of guppies, namely male coloration and pigmentation pattern ( Fig. 2 a, c–e), most of them being linked to sex chromosomes. This is different from P. wingei , which because of its more bright pigmentation became suspicious to John Endler and was an issue of debate since then. Nevertheless P. w i n g e i males also share many components of their pigmentation patterns with males from the two other species, suggesting that this is an ancestral trait. This is comparable to the situation in the genus Xiphophorus , were the (mostly sex chromosome linked) macromelanophore pattern can be found in similar expression phenotype in different species; or in the genus Poecilia where the male coloration is almost alike in the sailfin mollies.

A mentioned above the genetic difference between P. obscura from East Trinidad and P. reticulata from West Trinidad has been realized in numerous studies using allozyme, microsatellites and mitochondrial DNA sequences (Carvalho et al., 1991; Russell and Magurran, 2006; Alexander and Breden, 2004). The absence of obvious morphological criteria and the fact that hybridization between both forms occurs readily in the Turure River as a consequence of the Haskins transplant has led to the conclusion that both forms are rather in an early stage of speciation. Russell and Magurran (2006) noted that laboratory produced hybrids between P. reticulata from West Trinidad and P. o b s c u r a display male behavioral sterility and hybrid breakdown for embryo viability, brood size and sperm counts. This indicates some genetic separation which was interpreted as an initial stage rather than what can be expected from long separated species. The extent to which interspecific hybrids show features of hybrid breakdown can vary, however, and is not especially prominent in other poeciliid species ( Schartl, 2008). Moreover, interspecific hybrids can be produced in the laboratory of most poeciliids tested so far. Prezygotic isolation mechanisms are especially efficient to prevent interspecific hybridization of naturally co-occurring species of the same genus, but may not exist in allopatric species. Thus the Turure explant might have created an artificial situation of “sympatry” like enforced breeding of allopatric species in the laboratory. A similar situation exists for P. w i n g e i, which was not recognized as a species for a long time although strong genetic differentiation and differences in male display traits, body shape and coloration were clearly documented ( Alexander and Breden, 2004). The presence of hybrids in the Cumana area as a result of human activities showing the Y-chromosome dominated phenotype of P. w i n g e i but having the mitochondrial genome of P. reticulata might have confused the interpretation of the analyses.

Distribution. P. obscura occurs in freshwaters of the Oropuche River system, North-eastern Trinidad. The species was found (confirmed by DNA sequencing) in the headwaters and upper regions of the northern tributaries to the Oropuche River including the Quare River and La Seiva River. It was also found in the Matura River, the Salybia /Rio Seco Rivers and the L’Ebranche River, which are not tributaries to the Oropuche River but flow independently next north and south into the Atlantic Ocean at the West Coast of Trinidad. The southern and northern boarder of the range of the new species have not been determined exactly, but there are guppies in the Tompire River that flows north of the Matura River and in the Mission River, which flows into the Caribbean Sea (Reznick, pers. comm.). Further west on the northern coastline all the rivers that flow into the Caribbean Sea have no guppies until the Madamas River is reached (Reznick, pers. comm.). This river belongs to the northwestern province, the guppies of which - according to the molecular phylogeny - clearly are P. reticulata . No information is available for the situation south of the L’Ebranche.

In 1957 C.P. Haskins released 200 P. reticulata from the lower Guanapo River of the Caroni drainage (West Trinidad) into a site of the upper Turure ( Magurran, 2005), a northern tributary to the lower Quare River that flows into the lower Oropuche. These fish spread out from there downstream in the Turure ( Shaw et al., 1992) ( Magurran, 2005). We noted by presence of the P. reticulata mitochondrial haplotype (data not shown) that the invaders have already reached the main river and could threat the newly described species either by displacement or hybridization. Hybrids between P. reticulata from the Caroni system (West Trinidad) and P. obscura are easily produced in the laboratory ( Russell and Magurran, 2006) and a gene flow from the transplantation site into the native population of the Turure was already noted in a microsatellite study (after Russel 2004, cited in Magurran, 2005)). Due to this problem the dispersal of P. obscura in the main river of the Oropuche system including its southern tributaries is unclear.

Figures

FIGURE 4. CPD / SL / CPD ratios of P. obscura (symbols, population Rio Seco (RS), population Oropuche (OR )) and P. reticulata females (symbols, population Claxton Bay (CB) population Caroni Swamp (CS )).

FIGURE 6. Scanning electron microscopy images of female genital pores of (a) P. w i n g e i (Laguna de los Patos), (b) P. reticulata (Caroni Swamp), (c) P. obscura, (Upper Oropuche). Bar represents 200 mm.

FIGURE 7. Phylogramm of P. obscura, and P. reticulata from different localities and P. w i n g e i based on approx. 2200 bp of mitochondrial DNA sequences. 50 % majority rule consensus tree rooted on M. picta as outgroup. Average bootstrap values obtained using different types of analysis (see Materials and Methods) are indicated above the branches. For abbreviations of origins see table 1. Numbers in brackets indicate different individuals from the same locality.

FIGURE 8. Phylogramm of P. o b s c u r a and P. reticulata and P. w i n g e i from different localities based on approximately 800 bp of cytochrome b DNA sequences. 50 % majority rule consensus tree rooted on M. picta as outgroup. Average bootstrap values obtained using different types of analysis (see Materials and Methods) are indicated above the branches. For abbreviations of origins see table 1. Numbers in brackets indicate different individuals from the same locality.

FIGURE 2. P. obscura male (a) and female (b) from locus typicus, Oropuche River (OR). (c) P. obscura, population Rio Seco (RS); (d) P, reticulata, population Caroni Swamp (CS); (e) P. reticulata, population Rio Yaguaracual (RY); (f) P. c f “ wingei ” from El Cordon waterfall (EC); (g) P. w i n g e i from locus typicus, population Campoma (Ca); (h) P. w i n g e i, population Laguna de los Patos (LP).