RAPD markers indicate the occurrence of structured populations in a migratory freshwater fish species

Many factors have contributed to the destruction of fish habitats. Hydroelectric dams, water pollution and other environmental changes have resulted in the eradication of natural stocks. The aim of this study was to detect the genetic variation in Prochilodus marggravii from three collection sites in the area of influence of the Três Marias dam (MG) on the São Francisco river (Brazil), using the RAPD technique. The results obtained revealed that the fish in the downstream region nearest the dam have a higher similarity coefficient than those from the other sampling sites that may be related to differences in environmental characteristics in these regions. Additionaly, significant differences in the band frequencies were observed from one collection site to another. These both findings suggest the occurrence of a structured population and have important implications for the conservation of the genetic variability of distinct natural P. marggravii stocks.


Introduction
Prochilodontidae fish are widely distributed throughout South America and have been reported in almost all South American hydrographic basins (Mago-Leccia, 1972).
Thirteen species of the Prochilodus genus are recognized (Castro, 1990) and three species, P. marggravii, P. affinis, and P. vimboides, live in the São Francisco basin (Britski, 1988), which comprises an area equivalent to 7.4% of Brazil's territory (Paiva, 1982;1983).The two former species are migratory and are considered very important species for inland fishing in this river basin (Paiva, 1983).However, a dam built in 1960 on the main channel of the São Francisco river in the municipality of Três Marias (MG) prevents migratory fish from lower areas reaching most of the river's upper tributaries.There is no information on the effect of the construction of a new dam on the populational genetic structure of Neotropical migratory fish fauna.It is known that migratory fish require favorable conditions, determined by endogenous and exogenous factors, which guide migration and spawning.Free fish migration across a dam is often imparied and genetic variation and allelic frequency of migrating populations can be changed.Moreover, the environmental conditions prevailing in this region's hydrographic system have undergone profound changes both upstream and downstream from the dam (Sato and Osório, 1986).Drastic environmental conditions, caused mainly by reduced water flow, lower water temperature and oxygen rates just downstream from the dam, may be influencing some of the biological features of the living organisms in this area (Sato et al., 1995).Histological studies of gonads in P. marggravii have revealed reproductive disturbances in the specimens caught in the downstream area nearest the dam when compared to individuals from more distant regions (Sato et al., 1995).
Little is known about the population genetics of Prochilodus.Previous enzyme electrophoresis, used to quantify the genetic variations of P. lineatus collected from three sites in the Paraná river basin, detected no genetic divergences among the three samplings, suggesting a unique genetic pool of P. lineatus spread along the studied area (Revaldaves et al., 1997).
Another useful methodology to assess genetic variations in fish populations is the RAPD (Random Amplified Polymorphic DNA) technique, which is used in fishery management and conservation genetics of wild populations.Based on the amplification of genomic DNA by PCR (Polymerase Chain Reaction) with arbitrary nucleotide sequence primers, RAPD can detect high levels of DNA polymorphisms and can produce fine genetic markers (Williams et al., 1990;Welsh and McClelland, 1990).Nevertheless, RAPD analysis has some limitations that must be considered.It shows dominant inheritance and marker/marker homozygotes cannot be distinguished from marker/null heterozygotes.In addition, it is unable to assign bands to specific loci unless a previous pedigree analysis is performed.In applying this method, it is assumed that populations are under the Hardy-Weinberg equilibrium, that polymorphic bands segregate in the Mendelian way, and that marker alleles from different loci do not comigrate to the same position in the gel (D'Amato and Corach, 1996).
In order to assess the genetic variation of P. marggravii in the area of influence of the Três Marias dam on the São Francisco river, RAPD reactions were performed on samples from three collection sites.The results promise to be useful for the fishery management, aquaculture and stock conservation of this species.

Sampling sites
Adult individuals of Prochilodus marggravii were caught in November 1996, January 1997 and March 1997 at three sampling sites along the São Francisco river (Três Marias region).The first site (region A) comprises the proximate area below the dam (with inhospitable conditions caused by the dam's construction).The second location lies between an area approximately 10 km below the dam (Cascalheira) and the confluence of the Abaeté and São Francisco rivers (region B).The third area (region C) consists of a stretch of about 20 km below this confluence in the São Francisco river (Figure 1).Fifty-six individuals of P. marggravii were analyzed: 19 from region A, 20 from region B and 17 from region C.

DNA extraction, PCR and electrophoresis
Genomic DNA from liver tissue was sampled.Standard proteinase-K digestion was followed by phenol:chloroform extraction (Sambrook et al., 1989).A set of 10 decamer oligonucleotides (Operon Technologies) was used in this study as single primers in the Polymerase Chain Reaction (PCR) (Table 1).The amplification conditions, slightly modified, were based on Williams et al. (1990), and several tests were carried out to verify the best amplification conditions (DNA and MgCl 2 concentrations).PCR was performed in 50 mM KCl, 10 mM Tris pH 8.3, 2 mM MgCl 2 , 100 µM each of dNTPs (Amersham Pharmacia Biotech), 5 pmol of 10-base primer, 0.5 U of Taq DNA polymerase (Amersham Pharmacia Biotech), and 25 ng of genomic DNA in a final volume of 25 µL.Amplifications were made in an MJ Research thermal cycler, model PT100, programmed for an initial denaturation step, 4 min at 94 °C, followed by 45 cycles of 1 min at 94 °C, 1 min at 36 °C, and 2 min at 72 °C.The program was concluded with an additional 3 min at 72 °C step before cooling to 4 °C.One negative control (absence of template DNA) was included for each set of amplifications.The reaction products 20 Hatanaka and Galetti Jr.  were separated by electrophoresis on a 1.4% agarose gel stained with ethidium bromide and 8% polyacrylamide gel stained with 0.17% silver nitrate.In each case, electrophoretic runs were standardized for a comparative analysis of the band patterns obtained.

Data analysis
Individual RAPD patterns were compared within and between sampling sites.Only reproducible well-marked amplified fragments were scored.For each genotype, the presence and absence of fragments were recorded as 1 or 0, respectively.A pairwised comparison of banding patterns was evaluated by Jaccard's similarity coefficient (Jaccard, 1901), using the NTSYS-PC analysis software (Rohlf, 1993).Statistical analyses were performed in order to test the significance of similarity coefficients for each primer among the sampling sites, using the nonparametric Kruskal-Wallis test of the Biostat 2.0 software program (Ayres et al., 2000), which compares three or more size independent samples.Dunn's post-test of the Biostat 2.0 software program (Ayres et al., 2000) was also used to identify which sites show significant differences from each other.Fisher's Exact Test was performed using the Biostat 2.0 program (Ayres et al., 2000) to ascertain the existence of significant differences in band frequencies among the collections sites.

Results
Ten primers were tested prior to the analysis, with the amplification products ranging from 300 to 3000 bp.A low variable pattern was observed among P. marggravii individuals when the OPP-4 and OPP-18 primers were used, while the remaining primers produced polymorphic patterns (Table 1).No characteristic and/or diagnostic bands were found for any sampling site.All the primers examined produced reproducible RAPD fragment patterns.Figure 2 shows an amplification pattern obtained with the OPP-12 primer.Four of the 10 tested primers were chosen (OPP-7, OPP-10, OPP-12, OPP-15), based on profile sharpness and on adequate variation levels for statistical tests.A total of 31 loci were analyzed and the results of Fisher's Exact Test revealed five loci with significantly different frequencies among the sampling sites (Table 2).
The similarity coefficients of Jaccard (1901) within and between sampling sites are given in Table 3, which shows that the lowest value of the similarity coefficients within samples was obtained for individuals from region C (0.442), whereas regions A and B presented Jaccard's coefficient values of 0.636 and 0.604, respectively.The Kruskal-Wallis test revealed that the mean similarity coefficients of Jaccard for each primer among the sampling sites were statistically significant (KW = 7.538; p = 0.011).Pairwise comparisons (Dunn's Multiple Comparisons test) demonstrated that the differences observed between regions A and B and between B and C were not significant (p > 0.05), but a statistically significant difference was found between regions A and C (p < 0.05), indicating a considerable genetic discrepancy between these regions.Jaccard's coefficient used between sampling sites showed a higher value between A and B (Table 3).

Discussion
RAPD can be an efficient tool to differentiate geographically and genetically isolated populations, and has been used to verify the existence of locally adapted populations within a species that may have arisen either through genetic selection under different environmental conditions or as a result of genetic drift (Fuchs et al., 1998).
Since homozygotes cannot be distinguished from heterozygotes by the RAPD technique, the absence of amplification of a band in two genotypes does not necessarily represent genetic similarity between them.Jaccard's similarity coefficient, which excludes negative co-occurrences, has been widely employed for statistical analysis of RAPD markers (Duarte et al., 1999).In the present study, our statistical analyses showed considerable genetic variation among individuals of P. marggravii from collection sites A and C.
Populational genetic differentiation can be driven by ecological, evolutionary and historical factors.In Barbus neumayeri, genetic differentiation among sampling sites that presented different oxygen rates could represent the effects of selective pressure (Chapman et al., 1999).The well-developed homing instinct of salmonid fish seems to be a decisive factor leading to strong population subdivisions (Ryman, 1983).An evolutionary unit can be identified for each tributary, with particular genetic traits possibly related to local adaptation and/or to inbreeding.In Oncorhynchus nerka, genetic differences were found between two populations inhabiting regions with distinct en- Hatanaka and Galetti Jr.  vironmental conditions (Hendry et al., 2000).Furthermore, some river or lake systems contain metapopulations composed of distinct breeding units (Carvalho, 1993;Hansen and Loeschcke, 1994).
Although further studies are required to confirm the repeatability of the patterns observed, at least three hypotheses can be put forward to explain the possible genetic structure observed in the P. marggravii in the Três Marias region.
First, the fish downstream from the confluence of the Abaeté and São Francisco rivers (region C) comprise a panmictic population (panmictic population model).During the reproduction period, one fraction migrates towards the dam; the higher similarity coefficient observed in individuals downstream from the dam (region A) is the result of genetic drift (Figure 3a).
Second, the findings reported on here possibly reflect the homing behavior of P. marggravii individuals, similarly to wild trout populations (Ryman, 1983).In this case, animals from region C comprise a metapopulation with distinct reproductive populations, which historically complete their migration to different parts of this hydrographic system (Figure 3b).Genetic differences are determined by dif-ferent evolutionary pathways of breeding populations in each tributary, while the remaining genetic similarities represent the putative genetic pool of a common ancestor stock.
In a third hypothesis only a minor fraction of this metapopulation completes its migration towards the dam (region A) during the reproductive period, while the majority possibly migrate to locations with environmental conditions more favorable for reproduction (structured population model).Individuals from different populations present co-migrational behavior during the reproductive period (Figure 3c).
Competition for greater resources is very common in nature.Considering that the area downstream from the dam lacks the optimal conditions for reproduction, the most vigorous individuals are expected to take over the best habitats, establishing territories and leaving the less favorable habitats for the weaker individuals during fish migration (Krebs and Davies, 1996).Previous studies have demonstrated that specimens collected from region C, which offers more favorable reproductive conditions, are larger than the fish in regions A and B. At its confluence with the São Francisco, the Abaeté river supplies the former with higher water temperature, flow and oxygen rates than the conditions downstream from the dam (Sato et al., 1995).It is well known that hydroelectric dams create new obstacles for the natural dispersion of migrant fish, directly affecting their survival and reproduction by changing thermal and hydrodynamic conditions (Agostinho et al., 1992).
Findings similar to those observed here in P. marggravii have also been reported for other migratory fish species studied in the same area.In Brycon lundii, the presence of a characteristic band in 100% of specimens from regions A and B, albeit in only 27.3% of individuals from region C, also suggested the occurrence of structured populations (Wasko and Galetti, 2002), reinforcing the argument that the model of structured populations is probably the one best suited to illustrate the distribution of P. marggravii in the Três Marias region on the São Francisco river.
These ideas highlight an important aspect of the genetic conservation of this species.The practice of releasing cultured fish into the wild is widespread, but without a careful genetic analysis, it may damage the goals of preservation, leading to the homogenization of populations and decreasing species diversity (Taggart and Ferguson, 1986;Fritzner et al., 2001).Furthermore, it is important to recognize that different scenarios require different measures.Hence, the aim of conservation programs should be to develop an integrated strategy that conserves as much genetic diversity within the species as possible, and ensures the presence of utilizable fish resources (Hansen and Loeschcke, 1994).
Structured populations of a freshwater fish 23 The findings reported in this study nevertheless reveal important implications for the fishery management, aquaculture, and stock conservation of P. marggravii in the Três Marias region of the São Francisco river basin.The scenario, however, is not complete and further samplings from additional sites, mainly along large tributaries such as the Abaeté river, taken during different seasons of the year, should be analyzed.

Figure 1 -
Figure 1 -Map of the Três Marias region, showing the sampling sites.A: downstream from the dam, B: region between Cascalheira and the Abaeté river, and C: downstream from the confluence of the Abaeté and São Francisco rivers.

Structured populations of a freshwater fish 21 Figure 2
Figure 2 -RAPD patterns from Prochilodus marggravii using primer OPP-12 from three sampling sites in the São Francisco river, section A: individuals collected from region A, B: individuals from region B, and C: individuals from region C. Molecular weight marker ladder 1 kb is denoted in lane L.

Figure 3 -
Figure 3 -Genetic structure models for Prochilodus marggravii from the Três Marias region (São Francisco river).a: Panmictic model; b: Homing model; c: Structured population model.A: region A, B: region B, and C: region C. White and black squares indicate distinct reprodutive units; arrows indicate the migration routes.

Table 1 -
Primer sequences and amplification product characteristics.

Table 2 -
Characterization of amplification products, band size in base pairs (bp) and their relative frequencies in the A (fA), B (fB) and C (fC) sampling sites; and P-values of Fisher's Exact Test found for sites A-B; B-C and A-C.
*indicates significant difference; -indicates test could not be performed.