Environmental Biology of Fishes 61: 371–379, 2001. © 2001 Kluwer Academic Publishers. Printed in the Netherlands.

Habitats and littoral zone fish community structure of two natural lakes in southeast Brazil Volney Vono & Francisco A.R. Barbosa Graduate Program in Ecology, Conservation and Wildlife Management, Federal University of Minas Gerais, P.O. Box 486, 30161-970 Belo Horizonte, Minas Gerais, Brazil (e-mail: [email protected]) Received 21 February 2000

Accepted 29 January 2001

Key words: Middle Rio Doce basin, tropical lakes, habitat structure, Eucalyptus impact Synopsis The fish community and habitat structure in the littoral zone of lakes Pedra and Hortˆencia, middle Rio Doce Valley, Brazil, were investigated in three sampling periods from August 1992 to May 1993. A total of 9106 fishes were collected, including 11 species in Lake Hortˆencia and 12 species in Lake Pedra. Diversity was higher in Lake Hortˆencia, although total fish abundance was the same for both lakes, with no significant temporal differences. The length distribution of the majority of fish species was similar between lakes. Geophagus brasiliensis was the dominant species (number/biomass) in all habitats of the two lakes in all three sampling periods. A striking difference in total fish abundance in relation to habitat types, which we attribute to physical variables, especially the abundance of macrophytes was observed. Negative impacts attributable to the local Eucalyptus sp. plantations were not detected in the fish community structure of Lake Hortˆencia, which is probably due to the maintenance of the riparian vegetation contributing to similar physical environmental conditions in the littoral zone and high similarity in fish community structure between the two lakes. However, further studies on fish community structure and potential Eucalyptus impacts are suggested as necessary to improve conservation measures of the lakes and their fish faunas. Introduction The importance of the littoral zone for juvenile and adult fishes has been well documented in many different lake systems (e.g. Callow 1973, Lyons 1987, Werner & Hall 1988, Gelwick & Matthews 1990, Hinch et al. 1991, Rossier 1995); however, the majority of the studies have focused on temperate regions. In tropical systems, we understand far less about littoral zone community structure, fish species habitat preferences (Balon 1974, Penczak et al. 1994), and seasonal patterns of its use. We do know that littoral zones are critical nursery and foraging areas in tropical systems particularly in seasonally flooded areas (Lowe-McConnell 1975, Welcome 1985), and detailed studies on its fish fauna may provide higher resolution on community structure and interactions in this zone. The Rio Doce Valley lake system has great historical and ecological importance. Most of it is located

in one of the last Atlantic forest remnants in Brazil (Rio Doce State Park), surrounded by a region of accelerated industrialization and urbanization, among which Eucalyptus spp. plantations are particularly important within the lakes’ drainage areas. According to Tundisi et al. (1997), there is an important link between the terrestrial and aquatic ecosystems in the Rio Doce natural lakes. In fact, the littoral zone of these lakes may contribute significantly to their metabolism and the fish communities’ dynamics, as they possess complex and diverse habitat types, represented by differences in depth, substrate, and macrophyte density gradients. In this paper, we characterize the littoral zone habitats and fish community composition of two Rio Doce Valley lakes and identify habitat correlates of fish distribution. We selected two lakes that differ in catchment disturbance as a first step in examining the impacts of Eucalyptus spp. plantations on littoral zone structure and function.

372 Study areas The studied lakes, Pedra and Hortˆencia, are located in the Interplanaltic Depression in the middle Rio Doce Valley (42◦ –43◦ W, 19◦ –20◦ S), Minas Gerais State, Brazil (Figure 1). There are two distinct hydrological periods in the region: dry (April–October) and wet (November–March). During the study period, the highest precipitation was recorded in November of 1992 (258 mm) and the lowest (0.0 mm) in February of 1993. The water level of the two lakes varies with season, but rarely overflow to adjacent areas. In the 1950’s, several intensive anthropogenic activities started in the catchment basin of most lakes of the study area including Eucalyptus spp. plantations, resulting in clear-cutting of extensive areas of the original Atlantic forest (Paula et al. 1997). In lakes Pedra and Hortˆencia, the original catchment vegetation was replaced by Eucalyptus plantation ca. 25 years ago. The plantations were later abandoned, allowing for regeneration. In 1988, the regenerated vegetation

of Lake Hortˆencia was again removed, but a 10-m wide belt of natural vegetation was maintained along its shoreline. Differently, Lake Pedra has maintained its regenerated Atlantic forest during the last 25 years. The major morphometric characteristics of the two lakes are shown in Table 1. The littoral zones of the lakes are physically similar and mainly characterized by three distinct regions: high declivity with no macrophytes nor shoreline vegetation, medium declivity with medium macrophyte density and the presence of shoreline vegetation, and shallow regions with macrophytes and shoreline vegetation in variable densities. Materials and methods Sampling sites Samples were collected in August/September 1992, January 1993, and April/May 1993. Initially, three major littoral habitat zones according to the macrophyte density were distinguished. We selected

Figure 1. Map of the study area showing part of Rio Doce Valley lakes’ system, with study sites indicated.

373 Table 1. Morphometric characteristics of lakes Pedra and Hortência, middle Rio Doce, Minas Gerais State, Brazil. Characteristic

Lake Pedra

Lake Hortência

Catchment vegetation

Tropical Atlantic forest 15.43 33.84 8.0 20.5

Eucalyptus plantation 32.97 94.76 10.0 13.76

1534 1.102

3189 1.570

Area (ha) Catchment area (ha) Maximum depth (m) Catchment mean declivity (%) Shoreline length (m) Shoreline development index1

1 Shoreline development factor = S/[2(aπ )−1/2 ] where S = shoreline length and a = area.

three replicate sites according to each of the habitat zones, corresponding to: habitat 1 (null or low macrophyte density – 0 to 50 g m−2 ), habitat 2 (medium density – 51 to 250 g m−2 ) and habitat 3 (high density – >250 g m−2 ). This selection was made independently of other physical variables. Physical, chemical, and biological habitat variables Variables were measured in the central portion of each of the nine sites twice per sampling period, during the morning (8:00–12:00 h) and afternoon (14:00–19:00 h). Dissolved oxygen (% saturation), pH, conductivity (µS cm−1 ), water temperature (◦ C), and chlorophyll-a were measured at 20 cm depth. Chlorophyll-a measurements followed Lorenzen (1967). Other physical measures included water volume, maximum depth, and water transparency (Secchi disk). Macrophytes were collected from four 30 × 30-cm square samples per site. Densities (g m−2 ) were estimated after washing and oven drying for 48 hours at 70◦ C, and classified in three categories. We estimated the number of macrophyte species present in each habitat, and the number of its life forms following Payne (1986). Shoreline vegetation, represented by the dominant plant growth form, was classified in three categories: 1 (none), 2 (creeping grasses), and 3 (developed herbs or sedges). The substrate was categorized according to particle size as follows: clay (<0.005 mm), silt (0.005 to 0.05 mm), fine sand (0.05 to 0.25 mm), medium sand (0.25 to 0.85 mm), coarse sand (0.85 to 9.5 mm) and rock (>9.5 mm), and expressed as percentage of dry weight (% d.w.). The organic matter content in the

sediment (% d.w.) was determined for duplicate samples in each habitat, following Jackson (1967). The percentage of habitat shading (area protected against the sunlight) was qualitatively estimated in three categories: 1 (<30%), 2 (30 to 60%), and 3 (>60%). The amount of debris and litter present at the bottom was categorized as 1 (null or little), 2 (medium amount), and 3 (high amount). Habitat complexity was estimated using a combination of physical variables as follows: wood debris and litter coverage percentage, macrophyte density, number of macrophyte life forms, and substrate rock percentage. The values varied from 5 to 15 corresponding to the sum of the original categories (1 to 3), except for rock percentage, for which the values were transformed (1 = null, 2 = 1 to 30%, and 3 = >30%). Fish sampling Preliminary sampling showed that seining was an effective method for fish sampling in the littoral zone of the lakes (Vono 1992). Fishes were sampled by two seine hauls per site using a 18 m long × 1.6 m deep nylon seine (4.0 mm stretch mesh). One end of the seine was fixed at the shore, and using a motor boat at constant speed, the other end was run freely into the water in a 12-m diameter, thus enclosing a 56.5 m2 semi-circular area. A diver accompanied each seine haul and raised up the seine in case of rocks or debris on the substrate. Visual observations were also used to detect species not collected by seining. Collected fishes were preserved in 10% formalin and identified, counted, measured to the nearest cm (SL), and weighed in the laboratory. Fish community attributes were obtained for each habitat and included: richness, species diversity (Shannon Index), and fish density (number and biomass m−2 ). Size class frequency distributions for species were derived from cumulative data over the three sampling periods. Statistical analysis Fish diversity and number/biomass fish density between lakes were tested by the Student’s t test. The seasonal (among periods) and spatial (among habitat types) variation within each lake in fish densities, diversity and fish species richness were tested using repeated measures ANOVA procedure. This analysis was used as the measures were taken in the same sample units over time, that generates a dependency among

374 the samples (Crowder & Hand 1996). When significant differences were detected, a Tukey test to posteriori comparisons was done. For all the analysis, p ≤ 0.05 was used as the level of statistical significance. A principal component analysis (PCA) (Manly 1986, Ter Braak 1995) was performed to arrange habitat types in a reduced number of axes, using environmental and fish community structure factors. Variables that presented correlation up to 80% (shade percentage, transparency, complexity and medium sand percentage) were eliminated from the analysis through forward selection. The analysis considered the data of both lakes together, as the trends were similar when analyzed separately; moreover the lakes showed similar environmental and fish communities characteristics. All the statistical analysis were run on SAS (Institute Inc. 1989a,b). Results Environmental factors In general, water transparency, feophytin-a and organic matter were higher in Lake Hortˆencia, while pH and chlorophyll-a was higher in Lake Pedra. The other environmental characteristics showed no significant differences among lakes (Table 2). Macrophytes density did not differ significantly among sampling periods in either of the two lakes (p > 0.05). Six macrophyte species were recorded, distributed in three groups: emergent rooted (Eleocharis interstincta (Vahl), Eleocharis sp., Rhynchospora cyperoides (Sw) and Rhynchospora sp.); rooted floating leaves (Nymphaea sp.), and free floating (Ludwigia sp.). All species were found in both lakes, except Ludwigia sp., wich was restricted to Lake Hortˆencia. E. interstincta was the dominant species, colonizing mainly shallow waters (<1 m deep); Ludwigia sp. and Nymphaea sp. showed the lowest densities. Predominant sediment texture was sand-clay silt, corresponding to 89% and 78% respectively for the habitats of lakes Pedra and Hortˆencia. Most habitats in lake Pedra had creeping grass as dominant shoreline vegetation while in Lake Hortˆencia this dominance was of developed herbs. Fish communities A total of 9106 fishes were collected from the two lakes, representing 13 species in six families and

Table 2. Environmental characteristics (range) for the littoral habitats of lakes Pedra and Hortˆencia from August 1992 to May 1993. Character

Lake Pedra

Lake Hortˆencia

O2 (% saturation) pH Conductivity (µS cm−1 ) Sediment organic matter (% d.w.) Chlorophyll-a (µg l−1 ) Feophytin-a (µg l−1 ) Water temperature (◦ C) Secchi depth (m) Volume (m3 ) Maximum depth (m) Shade (%) Habitat complexity Wood debris Litter (%) Macrophyte density (g/m2 ) Macrophyte richness Main macrophyte species Main adjacent vegetation Main sediment texture

95.3–105.9 6.0–7.7 34–48 6.9–7.4

95.2–107.6 5.0–7.5 25–45 7.3–8.4

0–6.4 0–5.5 24.5–32.0 2.0–2.5 24.0–78.3 0.6–1.6 0–30 5–15 1–3 1–3 0–517

0–3.0 0–14.4 24.0–32.0 2.5–4.0 21.6–76.8 0.6–1.6 0–60 5–15 1–3 1–3 0–518

0–4 Eleocharis interstincta Creeping grasses

0–3 Eleocharis interstincta Developed herbs

Sand-clay silt

Sand-clay silt

four orders. Characoids was the dominant order and Characidae the dominant family. Geophagus brasiliensis (Cichlidae) was the dominant species in both lakes, which also showed the broadest spatial distribution. A total of 11 species were common to both lakes, Astyanax bimaculatus (Characidae) was exclusive to Lake Hortˆencia and Crenicichla lacustris (Cichlidae) was exclusive to Lake Pedra (Table 3). Species diversity was significantly higher (t = 10.21; p < 0.001) in Lake Hortˆencia, but numerical (t = 0.38; p = 0.70) and biomass density (t = −0.91; p = 0.36) did not vary between lakes (Table 4). In Lake Hortˆencia, species diversity (F = 11.18; p < 0.010) and biomass density (F = 5.16; p = 0.049) differed among habitats (Table 5). These variables were higher in habitats with highest macrophyte density. Species diversity also differed among periods (F = 6.22; p = 0.013). During the dry season (August/September), diversity was higher than the wet season (January). In Lake Pedra, only numerical density varied among habitats (F = 5.14; p = 0.049), where those characterized by highest macrophyte density showed higher abundance. Species diversity (F = 10.99; p < 0.01), species richness (F = 8.10; p < 0.01) and numerical

375 Table 3. Species list (with order and taxonomic position arranged after Lauder & Liem (1993)), total number of specimens (N) and fish common names, caught with seine in the littoral zone of lakes Pedra and Hortˆencia, Minas Gerais State – Brazil, from August 1992 to May 1993. Taxa

Order Characiformes Family Characidae Astyanax taeniatus (Jenyns, 1842) Astyanax bimaculatus (Linnaeus, 1785) Astyanax sp. Moenkhausia doceana (Steindachner, 1876) Oligosarcus solitarius (Menezes, 1987) Characidium sp. Family Erythrinidae Hoplias malabaricus (Bloch, 1794) Family Anostomidae Leporinus steindachneri (Eigenmann, 1907) Order Gymnotiformes Family Gymnotidae Gymnotus carapo (Linnaeus, 1758) Order Siluriformes Family Auchenipteridae Parauchenipterus striatulus (Steindachner, 1876) Order Perciformes Family Cichlidae Cichlasoma facetum (Jenyns, 1842) Crenicichla lacustris (Castelnau, 1855) Geophagus brasiliensis (Quoy & Gaimard, 1824)

Common name

N Lake Lake Pedra Hortência

Table 4. Fish community attributes for lakes Pedra and Hortˆencia during the sampling periods. N = Total number of specimens; B = Total biomass (g); S = Species richness (total number of species); H = Species diversity (Shannon Index); ND = Numerical mean density (number per m2 ); BD = Biomass mean density (gram per m2 ). BD

12

0.57

4725

6823

11

0.90

9106

16 528

13

—

1.45 (±0.70) 1.55 (±0.68) —

3.17 (±1.65) 2.24 (±0.42) —

4381

Hortência Total

239

Lambari

—

36

Lambari Chatinha

45 80

1001 5

Peixecachorro Canivete

19

4

20

13

Traira

8

9

Species diversity

Piau

6

14

Species richness

Sarapó

2

9

1

ND

9705

Pedra

155

3

H

N

Lambari

Cumbaca

S

Lake

B

Table 5. Results of repeated measures ANOVA procedure to fish biological variables of lakes Pedra and Hortˆencia. Variable

Number density Biomass density

Effect

Habitat Period Interaction Habitat Period Interaction Habitat Period Interaction Habitat Period Interaction

Lake Hortˆencia

Lake Pedra

F(4,12)

p

F(4,12)

p

11.183 6.226 0.635 4.276 3.301 0.727 3.113 0.941 0.509 5.163 0.277 0.385

0.009 0.013 0.646 0.070 0.072 0.589 0.118 0.417 0.729 0.049 0.762 0.814

1.783 10.999 0.620 1.797 8.106 2.066 5.143 7.581 1.640 3.386 4.309 1.583

0.246 0.001 0.656 0.244 0.005 0.148 0.049 0.007 0.227 0.103 0.061 0.241

Bold text highlights differences at 0.05 significance level. Riscadinho Bastiana Cará

150

67

22

—

3873

3325

density (F = 7.58; p < 0.01) differed among sampling periods. Species diversity and richness were higher during the dry season, while numerical density was higher during the wet season. There was no significant difference in any characteristic considering the interactions of both time and space effects for the two lakes (Table 5). In both lakes, the littoral zone fish communities were comprised mainly of small species (<10 cm SL) and juveniles of larger species. Length-frequency

distributions for most species were similar between the two lakes. All species were represented by juvenile and adult fishes in both lakes, except for P. striatulus, which were found as adults only. The number of specimens relative to time sampling showed no obvious relationship. Relationships between environmental factors and fish communities The PCA analysis reduced the environmental and biological variables in three principal components. Eigenvalues for these components, 4.62, 3.19 and 2.82 respectively, accounted together for 50.6% of total variance of the model (Table 6). In this analysis, values up to 0.60 were considered as high loading. The

376 Table 6. Loading of fish communities and environmental variables on the first three factors extracted by PCA. PC1 Fish community parameters Diversity (H ) Species richness Numerical density Biomass density Environmental variables O2 pH Conductivity Organic matter percentage Water temperature Volume Maximum depth Wood debris Litter percentage Shoreline vegetation Macrophyte density Macrophyte richness Clay percentage Silt percentage Fine sand percentage Coarse sand percentage Rock percentage Eigenvalue % total variance explained

PC2

and indicates the habitats with medium macrophyte density (habitat 2).

PC3

Discussion 0.70 0.65 0.46 0.66

−0.27 −0.06 0.61 0.37

0.27 0.10 −0.17 −0.09

0.40 −0.45 0.41 −0.06 −0.42 −0.47 −0.73 −0.59 0.00 −0.21 0.69 0.29 −0.26 0.32 0.50 −0.39 0.26 4.62 22.00

−0.43 0.45 −0.61 −0.52 0.58 0.37 0.06 −0.41 −0.34 0.07 0.48 −0.01 −0.51 −0.15 0.16 0.16 0.48 3.19 15.20

0.18 −0.26 0.41 −0.56 −0.40 −0.15 −0.01 0.02 −0.66 0.36 −0.09 −0.52 −0.42 −0.61 −0.36 0.62 0.20 2.82 13.43

Bold text highlights loading up to 0.60.

analysis showed high relationships between fish community structure and physical/biological variables such as macrophyte density positively and maximum depth and wood debris negatively. Component 1 reflects a link between these variables with fish communities factors, and can be described as the complexity gradient of the habitats. Fish numerical density and limnological variables, as conductivity and water temperature, contributed to the second component. The third component was related to substratum parameters such as organic matter, litter, silt, and coarse sand percentage. Fish communities parameters (fish species richness, diversity and numerical and biomass density) together with macrophyte density are close to each other in positive values of PC1. Maximum depth and wood debris were mainly in opposite position (Figure 2a). Habitats characterized by higher macrophyte density (habitat 3) are most concentrated in the right half of the ordination diagram, while those with low macrophyte density (habitat 1) are most in left half (Figure 2b). A transition zone between these two habitat types were observed,

Fish community structure Among the species captured there was a dominance of Ostariophysi, mostly characoids which is a common pattern in the neotropics (Lowe-McConnell 1987), and confirmed by other studies in the region (e.g. Sunaga & Verani 1997, Vieira 1994). The high dominance of G. brasiliensis may relate to a preference for lentic habitats (Travassos & Pinto 1958), omnivorous feeding habit (Sabino & Castro 1990, Vieira 1994), and fractional spawning (Lamas 1993). A high abundance of juvenile G. brasiliensis was recorded both in the summer (January) and autumn (April and May), suggesting two major recruitment periods, a demographic feature corroborated by Vieira (1994). The low abundance of juveniles in nonvegetated areas and its positive relationship with macrophyte density suggest that littoral habitats are important for refuge and feeding. Although Lake Hortˆencia presents an area twice that of Lake Pedra, has higher shoreline development, and approximately three times its catchment area, species richness and abundance were similar in both lakes. Lake Hortˆencia did show a high fish diversity due to a lower abundance of G. brasiliensis and higher abundance of Astyanax sp. The similar characteristics of the littoral zones, the proximity of the lakes, and their shared geological origin (Pflug 1969) suggest that similar colonization processes may have contributed to the fish community attributes. The similarity in biomass and numerical density among sampling periods may be caused by stability of physical and chemical conditions in the littoral zone, facilitated by low amounts of rain during the sampling periods and the resulting low water fluctuations. Numerical abundance did vary temporally at Lake Pedra which reflected changes in the number of G. brasiliensis juveniles following the two major recruitment periods. Habitat conditions and fish community structure In the limnetic zone of three other lakes of the region, Barbosa (1997) and Sabar´a (1994) found considerable abiotic variation over time, particularly in the thermal structure of the water column, which were related to

377

Figure 2. Ordination of environmental and fish communities’ variables (a) and habitat types position (b) according to PC1 and PC2 (AVE = adjacent vegetation, CLP = clay percentage, COM = conductivity, DOX = dissolved oxygen, FBD = fish biomass density, FND = fish numerical density, FSD = fish species diversity, FSP = fine sand percentage, FSR = fish species richness, LIP = litter percentage, MAD = maximum depth, MDE = macrophyte density, MRI = macrophyte richness, OMP = organic matter percentage, PH = pH, ROP = rock percentage, SIP = silt percentage, TSP = thick sand percentage, WAT = water temperature, WAV = water volume, WOD = wood debris).

distinct characteristics of the seasons. Despite welldefined dry and rainy seasons in the study area, temporal variation in physical and chemical variables, including macrophyte densities in the littoral zone of the lakes was small. Over the study period, there was a 40 cm water level fluctuation, not enough to inundate the shoreline vegetation. Significant spatial variations for most limnological factors were not recorded, despite differences in macrophyte density and substrate composition in the littoral habitats. Thus, the observed spatial relationship between environmental factors and richness, diversity, and total fish abundance can be attributed to the temporal and spatial stability of limnological variables in both lakes. The relationship between physical factors and fish community structure has been reported by several authors for many aquatic environments (e.g. Gorman & Karr 1978, Eadie & Keast 1984, Lobb III & Orth 1991, Power 1992, Benson & Magnuson 1992). In this study, relationships between macrophyte density, habitat complexity, and intrinsic community parameters suggest that macrophytes are a key factor in structuring the fish communities, since the other components apparently did not affect diversity, richness, or abundance. The importance of macrophytes to the structuring of littoral communities has been well-documented (Goulding 1980, Ara´ujo-Lima et al. 1986, Rosas & Odum 1987, Savino & Stein 1989, Sazima & Machado 1989, Bettoli et al. 1992). In lakes Pedra and Hortˆencia,

macrophyte beds may provide copious food resources (detritus, insect larvae, and zooplankton) and protection, thus resulting in less predation risk for the dominant species G. brasiliensis. Substrates with debris can support a greater biological diversity (Payne 1986) mainly through greater habitat availability for macroinvertebrates and algae (Foltz 1982). The present results however, showed a negative correlation between debris, maximum depth, and fish. These findings lead to a misinterpretation of the existing fish-habitat relationships, due basically to the interactions of the correlated variables. Principal component 1 partly indicates this correlation, composed mainly of habitat complexity and diversity and abundance of fishes. An explanation is that depth is a limiting factor for macrophyte distribution and growth, thus reducing habitat complexity. Furthermore, littoral habitats showing high declivity receive higher allochthonous inputs (litter and logs) from the trees closer to the shore. As a consequence, negative relationships between debris and fish communities is assigned to the high declivity of habitat in which this resource was found, associated with a low macrophyte density. Impacts from Eucalyptus plantations In the study region, the main anthropogenic impact is deforestation and conversion to Eucalyptus plantations. These plantations can affect most physical-chemical

378 and biological processes within a water body (Garman & Moring 1991, Hornbeck & Swank 1992). Although possessing different activities in their catchment basins, the common maintenance of a surrounding belt with natural riparian vegetation, as in the case of Lake Hortˆencia, is believed to provide similar physical conditions and a reasonable protection for the existing aquatic microhabitats, thus guaranteeing a stability of basic features such as habitat complexity, declivity, macrophyte density and sediment texture, in turn supporting similar fish community structures. Without this protection the likely release of toxic substances, mainly phenolic compounds, by the Eucalyptus plantations would affect negatively the fish communities, as suggested by Gehrke et al. (1993). Furthermore, Eucalyptus’ litter tends to be highly resistant to decomposition (Bunn 1988, Sabar´a 1994) and this would alter decomposition dynamics. Eucalyptus leaves and branches were not observed in the littoral zone of Lake Hortˆencia in the present study and, similarly, were not present in the allochthonous material previously studied by Sabar´a (1994), thus supporting the hypothesis that the riparian vegetation belt blocks this litter fraction. However, it must be pointed out that some questions related to the impacts of Eucalyptus plantations on the existing microhabitats of lakes Pedra and Hortˆencia and their fish fauna still remain to be pursued; aspects such as nutrient availability and energy transference through these particular communities were not considered in this study. As pointed out by Towns (1991) and Appelberg et al. (1993), these are important aspects in analyzing the impacts of Eucalyptus on aquatic organisms. Furthermore, in a study on the allochthonous contribution from regenerated forests to Lake Pedra, Sabar´a (1994) demonstrated that this material contains three times more soluble carbohydrates, lipids and crude protein than that from Eucalyptus plantations at Lake Hortˆencia. This author concluded that the distinctly different activities in the catchment areas were responsible for the differences in phytoplankton biomass and composition between the two lakes. A similar conclusion was reached by De L´opez et al. (1996) referring to differences in zooplankton density. In conclusion, the present study was able to demonstrate that complexity of existing habitat, particularly the presence of macrophytes, is a key factor structuring fish communities and that further studies considering more lakes of the region, and a monitoring program are necessary to clarify potential threats of Eucalyptus

plantations on aquatic diversity, especially on the littoral communities.

Acknowledgements This study was partly financed through grants from the U.S. Fish and Wildlife Service and CNPq. We are grateful to ACESITA Company for supporting the field sampling, particularly to Millˆor Godoy Sabar´a. We also thank to Anderson Latini for the improvement of statistical analysis and suggestions, F´abio Vieira for the discussions during the study and Douglas Andrew Yanega for the English review.

References cited Appelberg, M., B. Henrikson, L. Henrikson & M. Svedang. 1993. Biotic interactions within the littoral community of Swedish forest lakes during acidification. Ambio 22: 290–297. Ara´ujo-Lima, C.A.R.M., L.P.S. Portugal & E.G. Ferreira. 1986. Fish-macrophyte relationship in the Anavilhanas archipelago, a black water system in the central Amazon. J. Fish Biol. 29: 1–11. Balon, E.K. 1974. Fish production of a tropical ecosystem. pp. 248–748. In: E.K. Balon & A.G. Coche (ed.) Lake Kariba: A Man-Made Tropical Ecosystem in Central Africa, Monographiae Biologicae 24, Dr W. Junk Publisher, The Hague. Barbosa, F.A.R. 1997. The importance of diurnal cycles for the conservation and management of tropical waters: examples from the Rio Doce Valley lakes system. pp. 449–456. In: J.G. Tundisi & Y. Saijo (ed.) Limnological Studies on the Rio Doce Valley Lakes, Brazil, Brazilian Academy of Science, Rio de Janeiro. Benson, B.J. & J. Magnuson. 1992. Spatial heterogeneity of littoral fish assemblages in lakes: relation to species diversity and habitat structure. Can. J. Fish. Aquat. Sci. 49: 1493–1500. Bettoli, P.W., M.J. Maceina, R.L. Noble & R.K. Betsill. 1992. Piscivory in largemouth bass as a function of aquatic vegetation abundance. North Amer. J. Fish. Manag. 12: 509–516. Bunn, S.E. 1988. Processing leaf litter in a northern jarrah forest stream, Western Australia. I. Seasonal differences. Hydrobiol. 162: 201–210. Callow, P. 1973. The food of Ancyslus fluviatilis (Mull), a littoral stone-dwelling herbivore. Oecologia 13: 113–133. Crowder, M.J. & D.J. Hand. 1996. Analysis of repeated measures. Monographs on statistics and applied probability, volume 41, Chapman & Hall, London. 257 pp. De L´opez, C.M., V. Vono & P.M. Maia-Barbosa. 1996. Avaliac¸a˜ o da comunidade zooplanctˆonica na regi˜ao litorˆanea de dois lagos naturais no M´edio Rio Doce. Arquivo Brasileiro de Medicina Veterin´aria e Zootecnia 48: 141–149. Eadie, J. McA. & A. Keast. 1984. Resource heterogeneity and fish species diversity in lakes. Can. J. Zool. 62: 1689–1695.

379 Foltz, J.W. 1982. Fish species diversity and abundance in relation to stream habitat characteristics. Proc. Ann. Conf. Southeast. Assoc. Fish and Wildl. Agencies 36: 305–311. Garman, G.C. & J.R. Moring. 1991. Initial effects of deforestation on physical characteristics of a boreal river. Hydrobiol. 209: 29–37. Gehrke, P.C., M.B. Revell & A.W. Philbey. 1993. Effects of river red gum, Eucalyptus camaldulensis, litter on golden perch, Macquaria ambigua. J. Fish Biol. 43: 265–279. Gelwick, F.R. & W.J. Matthews. 1990. Temporal and spatial patterns in littoral-zone fish assemblages of a reservoir (Lake Texoma, Oklahoma-Texas, U.S.A). Env. Biol. Fish. 27: 107–120. Gorman, O.T. & J.R. Karr. 1978. Habitat structure and stream fish communities. Ecol. 59: 507–515. Goulding, M. 1980. The fishes and the forest, explorations in Amazonian natural history. University of California Press, Berkeley. 280 pp. Hinch, S.G., N.C. Collins & H.H. Harvey. 1991. Relative abundance of littoral zone fishes: biotic interactions, abiotic factors, and postglacial colonization. Ecol. 72: 1314–1324. Hornbeck, J.W & W.T. Swank. 1992. Watershed ecosystem analysis as a basis for multiple-use management of eastern forests. Ecol. Appl. 2: 238–247. Jackson, M.L. 1967. An´alisis qu´ımico de suelos. Ed. Omega, Barcelona. 662 pp. Lamas, I.R. 1993. An´alise de caracter´ısticas reprodutivas de peixes brasileiros de a´ gua doce, com eˆ nfase no local de desova. Dissertac¸a˜ o de Mestrado, UFMG, Belo Horizonte. 72 pp. Lauder, G.V. & K.F. Liem. 1983. The evolution and interrelationships in the actinopterygian fishes. Bull. Mus. Comp. Zool. 150: 95–197. Lobb III, M.D. & D.J. Orth. 1991. Habitat use by an assemblage of fish in a large warmwater stream. Trans. Amer. Fish. Soc. 120: 65–78. Lorenzen, C.J. 1967. Determination of chlorophyll and pheopigments: spectrophotometric equation. Limnol. Oceanogr. 12: 343–346. Lowe-McConnell, R.H. 1975. Fish communities in tropical freshwaters; their distribution, ecology and evolution. Longman, London. 337 pp. Lowe-McConnell, R.H. 1987. Ecological studies in tropical fish communities. Cambridge University Press, Cambridge. 381 pp. Lyons, J. 1987. Distribution, abundance, and mortality of small littoral-zone fishes in Sparking Lake, Wisconsin. Env. Biol. Fish. 18: 93–107. Manly, B.J.F. 1986. Multivariate statistical methods, a primer, third edition. Chapman & Hall, London. 179 pp. Paula, J.A. 1997. Biodiversidade, populac¸a˜ o e economia: uma regi˜ao de mata atlˆantica. UFMG/Cedeplar; ECMXC; PADCT/CIAMB, Belo Horizonte. 672 pp. Payne, A.L. 1986. The ecology of tropical lakes and rivers. John Wiley & Sons, New York. 301 pp. Penczak, T., A.A. Agostinho & E. K. Okada. 1994. Fish diversity and community structure in two small tributaries of the Paran´a River, Paran´a state, Brazil. Hydrobiol. 294: 243–251. Pflug, R. 1969. Quaternary lakes of eastern Brazil. Photogrammetria 24: 29–35.

Power, M.E. 1992. Habitat heterogeneity and the functional significance of fish in river food webs. Ecol. 73: 1675–1688. Rossier, O. 1995. Spatial and temporal separation of littoral zone fishes of Lake Geneva (Switzerland–France). Hydrobiol. 300/301: 321–327. Rosas, L.R. & W.E. Odum. 1988. Occupation of submerged aquatic vegetation by fishes: testing the roles of food and refuge. Oecol. 77: 101–106. Sabar´a, M.G. 1994. Avaliac¸a˜ o dos impactos do plantio de Eucalyptus spp. sobre dois lagos naturais no m´edio rio Doce-MG: propostas de mitigac¸a˜ o e manejo. Dissertac¸a˜ o de Mestrado, UFMG, Belo Horizonte. 156 pp. Sabino, J. & R.M.C. Castro. 1990. Alimentac¸a˜ o, per´ıodo de atividade e distribuic¸a˜ o espacial dos peixes de um riacho da floresta atlˆantica (sudeste do Brasil). Rev. Bras. Biol. 50: 23–36. SAS Institute Inc. 1989a. SAS/Stat User’s Guide, fourth edition, volume 2, Cary. 846 pp. SAS Institute Inc. 1989b. SAS/Stat User’s Guide, fourth edition, volume 1, Cary. 943 pp. Savino, J.F. & R.A. Stein. 1989. Behavior of fish predators and their prey: habitat choice between open water and dense vegetation. Env. Biol. Fish. 24: 287–293. Sazima, I & F.A. Machado. 1989. Melhor em seco que na a´ gua: uma t´atica defensiva do peixe Laetacara dorsigera (Cichlidae). Ciˆenc. e Cult. 41: 1014–1016. Sunaga, T. & J.R. Verani. 1997. The fish communities of four lakes. pp. 359–369. In: J.G. Tundisi & Y. Saijo (ed.) Limnological Studies on the Rio Doce Valley Lakes, Brazil, Brazilian Academy of Science, Rio de Janeiro. Ter Braak, C.J.F. 1995. Ordination. pp. 91–173. In: R.H.G. Jongman, C.J.F. Ter Braak & O.F.R. Van Tongeren (ed.) Data Analysis in Community and Landscape Ecology, Cambridge University Press, Cambridge. Towns, D.R. 1991. Ecology of leptocerid caddisfly larvae in an intermittent South Australian atream receiving Eucalyptus litter. Freshwat. Biol. 25: 117–129. Travassos, H. & S.Y. Pinto. 1958. Estudos sobre a fam´ılia Cichlidae (Perciformes, Actionopterygii). Bol. do Mus. Nac. 175: 1–9. Tundisi, J.G., Y. Saijo & T. Sunaga. 1997. Ecological effects of human activities in the middle Rio Doce lakes. pp. 477–482. In: J.G. Tundisi & Y. Saijo (ed.) Limnological Studies on the Rio Doce Valley Lakes, Brazil, Brazilian Academy of Science, Rio de Janeiro. Vieira, F. 1994. Estrutura de comunidade e aspectos da alimentac¸a˜ o e reproduc¸a˜ o dos peixes em dois lagos do m´edio Rio Doce, MG. Dissertac¸a˜ o de Mestrado, UFMG, Belo Horizonte. 76 pp. Vono, V. 1992. Utilizac¸a˜ o de dois m´etodos para captura de peixes em regi˜ao litorˆanea de sistemas aqu´aticos lˆenticos. pp. 104–112. In: H. P. Godinho & P. Maia-Barbosa (ed.) Encontro Anual de Aquicultura de Minas Gerais 10, Associac¸a´ o Mineira de Aquicultura, Belo Horizonte. Welcome, R.L. 1985. River fisheries. FAO Fisheries Technical Paper, Rome. 330 pp. Werner, E.E. & D.J. Hall. 1988. Ontogenetic habitat shifts in bluegill: the foraging rate-predation risk trade-off. Ecol. 69: 1352–1366.