Genetic evidence for two species of the genus Pimelodus Lacépède , 1803 ( Siluriformes , Pimelodidae ) in the Iguaçu River ( Brazil )

The existence of reproductive isolation between two morphs of catfish, endemic to the Iguaçu River (Brazil), was examined by enzyme starch gel electrophoresis. Tissues of 19 catfish (Pimelodus ortmanni) and 15 of a similar morph (Pimelodus sp.), which differs from P. ortmanni by presenting larger and more scattered dusky spots on its skin, were analyzed. A Nei’s (1978) genetic identity of 0.551 was determined by the analysis of 22 enzyme loci. The loci EST*1, EST*2, GDH*1, GPI*1, GPI*2, IDH*1, MDH*1, MDH*2, and PGM*1 were fixed for different alleles in each morph, that is, no heterozygote was found for these loci. The enzymatic patterns observed for the two morphs indicate both that the taxa are reproductively isolated and that they in fact represent separate species. Núcleo de Pesquisas em Limnologia, Ictiologia e Aqüicultura, Universidade Estadual de Maringá, Av. Colombo, 5790, 87020-900 Maringá, PR, Brasil. Send correspondence to E.R. Fax: +55-44-263-1424. E-mail: renesto@nupelia.uem.br


INTRODUCTION
The ichthyofauna of the Iguaçu River, a left bank tributary of the Paraná River, is highly endemic (Garavello et al., 1997) because of the Iguaçu Falls, a natural barrier limiting fish dispersal that has existed for about 22 million years (Severi and Cordeiro, 1994).The high declivity and velocity of its water makes the Iguaçu River basin useful for hydroelectric purposes, which has resulted in the construction of a series of dams along the river.This fact has led to changes in the characteristics of the river and has threatened some highly endemic fish species (Julio et al., 1997).
Until 1995, only one species of the genus Pimelodus had been known in the Iguaçu River (Pimelodus ortmanni Haseman, 1911).Garavello and Shibatta (1995) reported a new Pimelodus species.According to Garavello et al. (1997), Pimelodus sp. has a larger muzzle and a taller body than Pimelodus ortmanni.In addition, Pimelodus sp. has larger and more scattered dusky spots on its skin than does P. ortmanni (Figure 1).These authors distinguished the two Pimelodus species on the basis of these morphological differences.However, morphology is not always a sufficient basis for the recognition of separate species.Morphological polymorphism is observed among populations of polytypic species, whereas cryptic species are sometimes morphologically indistinguishable.
The electrophoresis of enzymes on starch or polyacrylamide gels provides a powerful test of the validity of presumed species.Because this technique allows the measurement of genetic distances among individuals, it can serve as a means for identifying similar species (Shaklee et al., 1982).This approach is particularly useful in cases of syntopy.In such situations, genetically differentiated species are easily recognized when fixed allelic differences are detected (Bernardi and Goswami, 1997).
We used enzyme horizontal starch gel electrophoresis to determine the level of genetic identity between the two morphs of Pimelodus from the Iguaçu River.

MATERIAL AND METHODS
Thirty-four fishes were collected from the Salto Caxias reservoir of the Iguaçu River using simple gill nets (Figure 2).The liver, white muscle and heart were removed from freshly caught fish and frozen in liquid nitrogen.The tissues were homogenized with plastic pestles in 1.5-ml Eppendorf tubes with four drops of 0.2 M Tris/HCl buffer, pH 7.5.Because of the large quantity of fat present in the liver, 0.5 ml of CCl 4 was added during homogenization (Pasteur et al., 1988).The homogenized samples were centrifuged at 44,720 g for 30 min at 1-5ºC.The supernatant fractions were analyzed by horizontal electrophoresis on 13% corn starch (Penetrose 30 ® ) covered with an ice pack.Two buffer systems were used: Tris-citrate, pH 7.0 (Shaw and Prasad, 1970) and Tris-borate-EDTA, pH 8.6 (Boyer et al., 1963).Electrophoresis was performed for 6 h at 5ºC with voltage gradients of 14 V/cm for the Tris-citrate gels, and 25 V/cm for the Tris-borate-EDTA gels (Table I).The gels were sliced horizontally and then incubated in the staining solution appropriate for each enzyme.
To visualize specific isozymes the procedures of Aebersold et al. (1987) were followed, except for aspartate aminotransferase (AAT) for which the method of Morizot and Schmidt (1990) was used.The nomenclature used is that proposed by Shaklee et al. (1990).The data were analyzed for Hardy-Weinberg equilibrium, allele frequency at each locus, frequency of polymorphic loci and Nei's genetic distance using the Biosys 1 software (Swofford and Selander, 1981).The genetic interpretation Genetic evidence for two species of the genus Pimelodus Lacépède, 1803 (Siluriformes, Pimelodidae) in the Iguaçu River (Brazil) Erasmo Renesto, Cláudio Henrique Zawadzki and Eloísa Revaldaves

Abstract
The existence of reproductive isolation between two morphs of catfish, endemic to the Iguaçu River (Brazil), was examined by enzyme starch gel electrophoresis.Tissues of 19 catfish (Pimelodus ortmanni) and 15 of a similar morph (Pimelodus sp.), which differs from P. ortmanni by presenting larger and more scattered dusky spots on its skin, were analyzed.A Nei's (1978) genetic identity of 0.551 was determined by the analysis of 22 enzyme loci.The loci EST*1, EST*2, GDH*1, GPI*1, GPI*2, IDH*1, MDH*1, MDH*2, and PGM*1 were fixed for different alleles in each morph, that is, no heterozygote was found for these loci.The enzymatic patterns observed for the two morphs indicate both that the taxa are reproductively isolated and that they in fact represent separate species.

S
of enzymatic patterns was based on the quaternary structure of the enzymes as described by Ward et al. (1992).

RESULTS
Twenty-two putative loci were detected based on 12 enzymatic systems analyzed for the 15 Pimelodus sp. and 19 P. ortmanni specimens (Table II).
The expression of the loci MDH*1 and PGM*1 (see Table I for abbreviations) was limited to P. ortmanni (Figure 3).
AAT: This dimeric enzyme showed greater activity in the liver than in heart or muscle.All the specimens gave two well-separated bands.This phenotype was interpreted as two loci without heterodimer formation.
ADH: This dimeric enzyme showed activity only in the liver.In P. ortmanni liver three phenotypes were seen: a single cathodic band, a single anodic band, and a mixture of three bands (one cathodic, one anodic and one at the origin).All the Pimelodus sp.specimens had only one cathodic band which migrated farther from the origin than the cathodic band of P. ortmanni.This pattern may be explained as a single locus with three alleles; Pimelodus sp. was monomorphic at this locus.
EST: This monomeric enzyme was restricted to the liver tissue and showed two anodic bands in both morphotypes, although those of Pimelodus sp. were less anodic than those of P. ortmanni.These phenotypes were assumed to result from two loci monomorphic for different alleles.
GDH: Pimelodus sp. had a single band which was more anodic than that of P. ortmanni.Both were considered monomorphic for different alleles at a single locus.
G3PDH: All but one of the specimens showed three bands for this enzyme in muscle.Since this enzyme has a dimeric structure, the three bands must result from two loci with heterodimer formation.Only one fish (Pimelodus sp.) gave five bands; it was likely heterozygous for two alleles at a single locus.
GPI: All of the specimens showed three bands, but the three bands of Pimelodus sp. were less anodic than those of P. ortmanni.This dimeric enzyme must be codified by two loci with heterodimer formation.The two morphotypes were monomorphic for different alleles.
IDHP: Both liver and heart gave a single band for this activity.The heart enzymes in the two morphotypes were coincident, but those of the liver differed.The liver enzyme of Pimelodus sp.migrated faster than the heart enzymes and that from P. ortmanni liver.Only one specimen of P. ortmanni gave three bands in heart tissues.The monomorphic active locus in the liver is therefore different from that in the heart.The three-band phenotype in heart samples is indicative of heterozygosity for two alleles.
LDH: All of the specimens gave five bands.Since this enzyme has a tetrameric structure, the phenotype must result from two loci which are monomorphic in both morphotypes.MDH: For this enzyme, Pimelodus sp.exhibited two bands and P. ortmanni had four.These phenotypes may be explained by two loci without heterodimer formation in the former morphotype and by three loci with heterodimer formation between the more anodic forms in the latter morphotype.Only locus MDH*3 of Pimelodus sp. was polymorphic.

SALTO CAXIAS DAM
MEP: The expression of this enzyme in liver was different from that in the heart in both morphotypes.Both of the tissues gave four bands.In the heart the most intense band was the most anodic and in liver the most intense band was the least anodic.These phenotypes were interpreted as resulting from two monomorphic loci.
PGM: A single band in Pimelodus sp. and two bands in P. ortmanni were detected for this enzyme.In both morphotypes there were secondary bands.These patterns indicate a single monomorphic locus in the former morphotype and two loci in the latter.
SOD: All specimens gave a single band indicative of a single locus.
P. ortmanni was polymorphic only at ADH*1 and IDH*2, and Pimelodus sp.only at G3P*1 and MDH*3.Moreover, the two species were fixed for different alleles at nine of the 22 loci (Table II).The unbiased estimated mean heterozygosity per locus (Nei, 1978) was 0.006 for Pimelodus sp. and 0.024 for P. ortmanni.All polymorphic loci assayed were in Hardy-Weinberg equilibrium.Nei's unbiased genetic identity and unbiased genetic distance (Nei, 1978) between the two species was 0.551 and 0.597, respectively.

DISCUSSION
The biological species concept is based on the reproductive isolation of groups of true-breeding populations from other such groups (Mayr, 1970).From conventional definitions under the biological species concept, two syntopic, conspecific morphs should have the same gene frequencies at each locus (Thorpe and Solé-Cava, 1994).I for enzyme abbreviations.
If two morphs or populations share no common allele at a locus, they must be reproductively isolated and considered true biological species.
The electrophoresis data revealed a low genetic diversity in the two morphs of Pimelodus.The heterozygosities estimated for Pimelodus sp.(0.006) and for P. ortmanni (0.024) were both below that estimated for Pimelodus maculatus (0.06) from the Tibagi River (another river in the Paraná River basin) by Almeida and Sodré (1998) and below the overall mean of 0.051 estimated for fishes by Ward et al. (1992).
The two morphs were fixed for different alleles at nine loci.No heterozygote was found for these loci.If the morphs were conspecific, a few heterozygotes should have been encountered in the population.Since 15 Pimelodus sp. and 19 P. ortmanni were analyzed, 68 alleles for each locus were screened.Thus, the expected allele frequencies would be p = 30/68 = 0.44 for the alleles of Pimelodus sp. and q = 38/68 = 0.56 for the alleles of P. ortmanni.The expected proportion of heterozygotes would therefore be 2pq = 2 x 0.44 x 56 = 0.493 for each locus.The probability of not obtaining a heterozygote for one of the seven loci would be 1 -0.493 = 0.507.The probability of not obtaining any heterozygotes in a sample of 34 individuals would be (0.507) 34 .For nine loci, this probability would be (0.507) 34x9 = (0.507) 306 .In other words, it would be impossible not to obtain any heterozygous individual for the nine loci.The null hypothesis of a single random breeding population is therefore rejected.
Based on published data on the frequency of differ-ent values of Nei's (1972) I for different levels of systematic divergence, Thorpe (1982) calculated empirical curves for probability against I values between confamilial genera, congeneric species and conspecific populations and concluded that if populations of uncertain status have genetic identities between 0.35 and 0.85, they should not be considered conspecific.The genetic identity between Pimelodus sp. and P. ortmanni was estimated to be 0.551, which is below the critical value of 0.85 established by Thorpe (1982).Thus, Pimelodus sp. could be considered a species that is actually distinct from P. ortmanni.
Taxonomically and ecologically, the Iguaçu River is still a poorly studied fluvial system.As a result, little is known of how these species were formed and why they are endemic to this basin.However, recent studies have provided some information on this subject.While working on the Iguaçu River around the Segredo reservoir (250 km upstream the Caxias dam, not shown in Figure 2), Suzuki and Agostinho (1997) noted intensive reproductive activity of Pimelodus sp.downstream from the Segredo dam, as well as massive reproductive activity of P. ortmanni in the Iratim River (a tributary of the Segredo reservoir, 60 km upstream the Segredo dam).This could mean that the two species require different reproductive sites, which could have led to the extensive genetic differentiation found between them.The recognition of the two populations and their presumed different reproductive sites needs to be taken into account by the Brazilian authorities when constructing new hydroelectric dams in the Iguaçu River basin to guarantee the genetic diversity of these fishes and their ecological importance.

ACKNOWLEDGMENTS
The authors thank COPEL (Companhia Paranaense de Energia Elétrica) for financial support, Dr. Oscar Shibata and Carla Pavanelli for the suggestions to the manuscript, Celso Ikeda for taking the photos, and Harumi I. Suzuki for collecting the fish samples.

Table II -
Allele frequency estimates for Pimelodus ortmanni and Pimelodus sp. from the Iguaçu River.
*See Table I for abbreviations.