DISTRIBUTION PATTERN OF TWO FISH SPECIES IN A COASTAL STREAM IN SOUTHEAST BRAZIL

Longitudinal distribution of Mimagoniates microlepis and Hoplias malabaricus was analyzed in a coastal stream system in Southeast Brazil. Six sites were sampled by electrofishing in seven months, from July 94 to July 95. Densities of M. microlepis, and H. malabaricus and its prey ( Poecilia vivipara, Deuterodon sp. and Astyanax janeiroensis ) were estimated by the Zippin method; environmental variables of: (i) kind of substratum; (ii) percentage of pools, runs, riffles, instream vegetation, and canopy were registered for each site in each sampling month. The relative importance of each biotic and abiotic variable was tested through simple correlation analyses. We verified that M. microlepis occurrence correlates with canopy; H. malabaricus is correlated with instream vegetation, pools, clay substratum, and P. vivipara densities; and P. vivipara correlates with pools and clay substratum. Our results suggest that environmental variables, and food and shelter availability are the main factors in determining M. microlepis and H. malabaricus distribution.


INTRODUCTION
Predicting parameters involved in organization of populations and communities continues to be a major challenge in animal ecology (Van Winkle et al., 1991;Matthews et al., 1994;Matthews, 1998).In trying to interpret mechanisms and processes underlying abundance and distribution of animal populations, divergent views have frequently developed throughout the history of ecology.Disagreement between defenders of biotic parameters, such as competition (e.g., Huntchinson, 1958;MacArthur, 1972), and abiotic ones, such as environmental settings (e.g., Andrewartha & Birch, 1954;Connor & Simberloff, 1979) has been common.This conflict continues, indicating the importance of physical versus biological processes in regulating the structure of stream-fish population and communities (Schlosser, 1987;Gelwick & Matthews, 1993;Flecker, 1997).
Some researchers emphasize the importance of physical variables (Grossman et al., 1982) and others have documented the importance of biological interactions (Fraser & Cerri, 1982;Power & Matthews, 1983).Bayley & Li (1992) suggest that environmental resources are the main factors in explaining fish species distribution and maintenance.Following this, the maintenance of populations reflects the interaction between environmental settings and adaptive characteristics of individuals of a species.Discussions go on, but present evidences corroborates both points of view.
In this paper we show the relative importance of environmental settings in determining densities and longitudinal distribution of Mimagoniates microlepis (Steindachner, 1876) and Hoplias malabaricus (Bloch, 1894) in the Ubatiba River.Behavioural and feeding habits of both species, provide the framework for these analyses.

Study area
The Ubatiba River (20 o S and 42 o W) together with its five tributaries composes a stream system in the Southeast of Brazil (Fig. 1).It flows for about 16 km, on the east side of Serra do Mar, and discharges in the Maricá Lagoon located around 70 km from Rio de Janeiro City.Deforestation, as a consequence of cattle breeding, is the main lowland characteristic.Nonetheless, preserved areas of Atlantic Forest are still common on the slopes and tops of the surrounding rocky hills.
The Ubatiba River system is composed of streams with clear water and its hydric regime fluctuates according to rainfall (~1,500 mm/year); more information about physicochemical characteristics of this system are presented elsewhere (Mazzoni, 1998;Mazzoni et al., 2000;Mazzoni & Lobón-Cerviá, 2000).

Sample procedure and data analysis
Six sites were sampled bimonthly in the Ubatiba River system (Fig. 1) between July 94 and July 95.These sites, between 70 and 100 m long, were selected to include all the variability of each particular stream.Width, depth, area, and water volume of each site, were determined at each sampling using bathymetric maps based on five-meter transects from the lowest sampled point downstream.Environmental settings were quantified by the percentage of silt, sand, pebble, gravel, and cobble; the sequence of pool, riffle, and runs was determined at each site.The relative importance of riparian and instream vegetation was considered as a percentage of the sampled area (m 2 , %) according to bathymetric maps (Table 1).
Fish numbers for each sampling site and date were determined by electrofishing techniques (220 V, 2.5 Amp, AC; see Mazzoni et al., 2000) through the three-removal method (Zippin, 1958), and transformed into fish density values for each sampled area.All collected fish were identified, measured (standard length, SL = mm), and released back into the water at a midpoint of the sampling site.All analyses were done according to mean fish densities at each site during the seven sampling periods.
To establish whether environmental settings determine density and distribution of M. microlepis and H. malabaricus, we used simple correlation analyses between their local mean densities and environmental settings; simple correlation analyses between mean densities of H. malabaricus and their prey (i.e., Poecilia vivipara Schneider, 1801, Astyanax janeiroensis Eigenmann, 1890 and Deuterodon sp.) were also done.

RESULTS
Descriptions of Ubatiba sites (Table 1) according to environmental settings show all presenting a combination of pools, runs, riffles, and variable substratum with no obvious longitudinal gradient but rather a mosaic of patches along the river; the negative correlation (p < 0.01) between canopy and instream vegetation is noteworthy.
The highest percentages of instream vegetation were registered at Ca, U2, and U5 while at Si, It, and U4 canopy was the predominant vegetation.
Sites U2 and Ca had no canopy and presented lower density of M. microlepis; in these cases, occur-rences were random and limited to 1 or 2 specimens.Even at site U5, with a poor canopy, the species was registered at the only shaded area (authors' personal observation).Sites Si, It, and U4 (28.4,25.0, and 12.7 ind/ha, respectively) presented high M. microlepis densities and canopy (Fig. 2).Regression analyses between M. microlepis densities and canopy showed a positive correlation (r 2 = 0.87; p < 0.002), the only significant one for M. microlepis.

DISCUSSION
Habitat use by fish species is related to morphological and physiological characters (Sale, 1969) including size, and form and position of fins in the first case (Wootton, 1990;Reilly & Wainwright, 1994), and trophic (Ringler, 1983) and reproductive requirements (Balon, 1975) in the second.
M. microlepis shows diurnal behaviour, preferring shallow waters with medium to slow water velocity.Allochthonous insects are the main food of species that swim through the surface and/or midwater to eat (Sabino & Castro, 1990).M. microlepis from the Ubatiba River feed mainly on terrestrial forms of Hymenoptera and Diptera (unpublished data), corroborating observations made by Sabino & Castro (1990).
Our results point to positive correlation between canopy and M. microlepis densities; the first attempt to explain this was based on the premise that covered areas are richer in allochthonous food, mainly terrestrial insects, than are open ones.Input of allochthonous matter is considered the most important source of food supply in streams (Welcomme, 1985;Luiz et al., 1998), compensating low primary production, a situation intensified in sites with intense vegetative cover because of reduced sunlight incidence (Power, 1984;Sechnick et al., 1986).Nevertheless, data in the literature suggest that there are no quantifiable differences of allochthonous input between covered and opened areas (e.g., Angermeier & Karr, 1983;Uieda & Kikuchi, 1995).In view of these contrasting results, we suggest that preference for shaded sites may be explained by both input of allochthonous food and protection from terrestrial predators, mainly piscivorous birds.A changing and heterogeneous background of sun and shade, as a consequence of sun flecks or dappling due to partially open canopy, has been registered as a way of decreasing predation in streams (e.g., Helfman, 1981;Sechnick et al., 1986).Based on this rationale, the abundance of M. microlepis in shaded areas may be explained by its predation avoidance and trophic requirements.Distribution of H. malabaricus can be approached in the same way.As a piscivorous species (Goulding, 1980;Lowe-McConnell, 1991), its main preys are A. janeiroensis, Deuterodon sp., and P. vivipara, the latter being most frequent in its diet (unpublished data).Our results indicate a strong correlation between H. malabaricus and instream vegetation and pool; the use of such an environment for forraging has been registered at a length for H. malabaricus (i.e., Uieda, 1984;Castro & Casatti, 1997).
The positive correlations recorded between P. vivipara and H. malabaricus densities, as well as for P. vivipara densities, pools, and instream vegetation are in agreement with some data in the literature.Caramaschi (1979) noted that pools and instream vegetation were the main habitat used by Phalloceros caudimaculatus (Hensel, 1868) (Poeciliidae related to P. vivipara) and young H. malabaricus which in this case used Poeciliidae as its main food item.Such results associated with ours, suggest that occurrence and abundance of H. malabaricus may be strongly related to available food, in this case P. vivipara.
Feeding habits of Poeciliidae fishes have been extensively discussed and all agree on its characterization as herbivorous/algivorous (Costa, 1987;Teixeira, 1989;Sabino & Castro, 1990;Aranha & Caramaschi, 1999).Lentic habitats with poor or no shade facilitate filamentous and unicellular algae bank formation.Therefore, positive correlation between P. vivipara and pools in open sites suggests causal relationships with feeding habits; moreover, the size of these individuals facilitate high density in these sites.
We conclude that occurrence and abundance of fish species of the Ubatiba system can be related to their trophic (selecting microhabitats so as to maximize their ability to efficiently use local food supplies) and anti-predator requirements.M. microlepis, a water column species, had its highest density registered at shaded sites, which increase food availability (i.e., terrestrial insects) and allows predation evasion.H. malabaricus and P. vivipara densities were correlated with pools and instream vegetation; such environments are highly propitious for feeding Poeciliidae which, in turn, is the main food item for H. malabaricus of the Ubatiba River system.
Finally, we suggest that the environmental variables studied are the main factors in the fish community structure in the Ubatiba system; nevertheless, further analysis on the other fish species are needed to corroborate our proposition.

Fig. 1 -
Fig. 1 -Geographical location of the Ubatiba River system showing the six sampling sites.

Fig. 2 -Fig. 3 -
Fig. 2 -Mean densities (ind/ha) of Mimagoniates microlepis and percentage of canopy in the six study sites in the Ubatiba River system.