The higher CPE was recorded in the Caracu that had higher sampling sites (3 vs. 2 in the São Pedro). Nevertheless, even the medium CPE by stream (5,906 vs. 3,897, respectively) was higher in the Caracu. It should be related to the two lentic environments made by human activity in the Caracu, where typical species such as Cichlasoma paranaense with parental care, and Phalloceros caudimaculatus with internal fertilization were always collected together with numerous juveniles.
The smaller water volume, as well as the oligotrophic conditions at the headwater might explain the lower CPE and the smaller number of species resident in relation to the subsequent downstream locations, as found by Vannote et al. (1980) for temperate rivers.
Headwaters. - These environments had a small volume of water, which explained why only a few resident species were present. Astyanax scabripinnis was constant and restricted to the headwater of the São Pedro. Besides the fact that this species is typical of headwater environments (Britski, 1972; Caramaschi, 1986), the presence of breeding individuals at this locality (Pavanelli, 1994) suggested that it was a resident. The Caracu headwater was considered a lentic environment, which explained the constancy of species that prefer low current flow, like Hoplias aff. malabaricus and Phalloceros caudimaculatus. Physical conditions in the two stream headwaters differed from each other because of human activity, which altered a lotic to a lentic environment in the Caracu. Lentic environments are not typical of headwaters. This fact led to faunistic differences, primarily regarding abundance and constancy.
Intermediary locations. - The Caracu intermediary sampling site was semi-lotic, with some grasses along the banks. In spite of this, Gymnotus carapo, Hoplias aff. malabaricus, and Phalloceros caudimaculatus, which have been found as characteristic of lentic environments (Britski, 1972), were recorded as constants. Additional constants were Cichlasoma paranaense, generally associated with macrophytes in low-flow environments (Kullander, 1983), and Astyanax altiparanae, which had been very common and widely distributed in Paraná basin streams (Britski, 1972). It could be interesting to mention that the stream middle stretch had more species than the headwater, as also noted by Lowe-McConnell (1975) and Mosley (1987).
Mouths. - In both streams, these sampling sites showed the highest richness and number of constant species. Many of the accessory and occasional species were represented by young stages of large species, typical of the Rio Paraná, which remained there for a short time, primarily during the recruitment season. Only four constant species, A. altiparanae, Serrapinnus notomelas, Hyphessobrycon eques, and Hypostomus ancistroides, were common to the mouths of both streams. These were resident species and apparently constituted the typical ichthyofauna of this stretch of the streams of that region.
The differences demonstrated at the mouths, primarily in species richness, were attributable to the grasses near the Caracu's mouth. Vegetation occurred further upstream in the São Pedro, reducing its attractiveness to young individuals from the Paraná seeking food and shelter. Thus the results allowed us to infer that the Caracu had higher availability of microhabitats for small fishes at its mouth than the São Pedro.
Penczak et al. (1994), working in the Caracu and another stream of the same region, used other fishing gears (basically electrofishing), and different data analysis. Their results are not comparable because of the selectivity of the fishing gear employed.
The cumulative frequency data for the stream mouths did not reach an asymptote indicating that some rare species were not been collected and/or that there was some fluctuation in the environment, showing its temporary character. The contribution of the Rio Paraná ichthyofauna, composed of the highest number of species, permits us to predict future occurrences of the river fauna in this area.
Margalef (1986) discussed measurements of spatial diversity, representing a diversity spectrum. This could form different patterns for natural populations. The most common pattern was the diagonal, which indicated heterogeneity along the environment analyzed, and the rectangular, which suggested that the samples were formed basically by the same species in the same proportions. The results of this study showed a diagonal pattern, where the ichthyofauna diversity increased from the headwater to the mouth in both streams. Lowe-McConnell (1975; 1987) and Gorman and Karr (1978) suggested that the stream's upper regions are relatively homogeneous environments, with lower habitat complexity and species numbers. Thus, in areas closer to the mouth, the environment becomes complex, offering physical conditions for sheltering a larger number of species. Moreover, the proximity to the Rio Paraná led to faunal mixing, where the main channel contributed to the increase in species diversity.
In the Caracu, there was a slight decrease in the diversity index in the receding and dry water seasons, which, like evenness, increased during the rest of the year. The little peaks were caused by increase in the species number in these phases of the hydrological cycle, when there was greater environmental heterogeneity, suggesting that these parameters were influenced by the hydrological regime.
In the São Pedro, the diversity index and evenness tended to rise during the sampling period. During receding and dry water seasons, some of the more numerous species decreased in abundance, causing an increase in evenness. Since the spectrum shown was a cumulative measurement, it was expected that there would be an increase in species richness and a consequent increase in species diversity during the sampling period.
The correspondence analysis of the sites located in the same region of the streams suggested that the distribution of the ichthyofauna obeyed a spatial gradient. The larger similarity between the stream mouths confirmed the influence of the Rio Paraná fauna. The physical and chemical variables of the water also showed a spatial gradient along the streams (Pavanelli et al., 1997).
The differences in CPE between the intermediary site and the mouth of the São Pedro could be explained by the fact that the mouth, in addition to being more influenced by the Paraná ichthyofauna, had a higher volume of water.
Intermediary. - This site had three constant species, A. altiparanae, Leporinus friderici, and Prochilodus lineatus, all common and typical of the Rio Paraná basin rivers. Among the accessory and occasional species were some typical of large rivers such as Brycon orbignyanus, the youngs or adults, which occasionally traveled upstream to this site, which was closer to the mouth and had lotic characteristics.
Mouth. - Net samples from the stream mouth showed similar constancy results to those from the sieves, however, with larger species.
Paraná (based on Nupélia/Finep, 1989). - The Rio Paraná had 9 constant, 13 accessory and 34 occasional species, including small, medium and large-sized species. Of the species belonging to the Rio Paraná ichthyofauna, 18 occurred only at the São Pedro mouth, showing the Rio Paraná contribution to the stream, with sporadic or even persistent species.
Only the São Pedro intermediary station reached an asymptote for species richness, as was also established using sieves. The results from the mouth showed that more species should occur in the lower part. It was notable that the Rio Paraná had 24 species that did not occur at any sampling site in the stream.
The spectrum of spatial diversity showed a diagonal pattern like that of the sieve samples, and was explainable by the difference in the environments analyzed. Similarly, the influence, as well as the higher availability of the Paraná microhabitats was confirmed in the lower parts.
The spectrum of temporal diversity indicated a low degree of influence by the hydrological regime, with a plateau beginning in April. Nets were mostly selective for large specimens and/or fishes of the Rio Paraná fauna, which belonged to migrating species that replaced each other proportionately, while the diversity did not change sharply. This indicated that the greatest influence came from the Rio Paraná fauna, and not from the flood regime.
The Sørensen index, in spite of yielding values quite close together, showed that the most similar sites were the mouth of the São Pedro and the Rio Paraná, which tended to confirm the mixture of faunas.
In summary, the ichthyofauna distribution in the streams analyzed was strongly influenced by the Rio Paraná at the spatial level, mainly in the lower parts, and by the river's hydrological regime at the temporal level. Although some species were constants, occasionals predominated. Corresponding sites in the two streams fell into groups of higher similarity. In addition to resident species, even among the more abundant or frequent ones, there were transitory individuals from the Rio Paraná fauna.
We are grateful, from UFRJ, to R. I. Rios and, from Nupélia, to A. A. Agostinho, S. M. Thomaz, S. Veríssimo, L. C. Gomes and N. S. Hahn for the critical reading of the manuscript, to R. Fugi for determining the sex and gonadal stages of the fishes, to L. M. Bini for helping in the correspondence analysis, to J. L. L. Pereira for drawing the Fig. 1, and to G. Baumgartner, E. Gonçalves, G. Gumieri, S. Rodrigues, S. Veríssimo, and T. A. Paggioro for helping in the hard fieldwork. Logistic support came from Nupélia. EPC and CSP are also gratefully acknowledged for grants from CNPq.