IMPACTS OFA RESERVOIR ON FISH ASSEMBLAGES OF SMALL TRIBUTARIES OF THE CORUMBA ´ RIVER, BRAZIL TADEUSZ PENCZAK, a * ANGELO A. AGOSTINHO, b LUIZ C. GOMES b and JOA ˜ O D. LATINI b a Department of Ecology and Vertebrate Zoology, University of Lo ´dz ´, Lo ´dz ´, Poland b Nu ´cleo de Pesquisas em Limnologia, Ictiologia e Aquicultura (Nupelia), Departamento de Biologia, Universidade Estadual de Maringa ´, Maringa ´, Parana ´, Brazil ABSTRACT Fish community surveys were conducted in five tributaries of the Corumba ´ River before and after damming. Electrofishing samples were collected monthly in the pre-impoundment period (March 1996 to August 1996) and 15 in the post-impoundment period (September 1996 to February 1999). A self-organizing map (SOM, an Artificial Neural Network algorithm) was used to represent the patterns of fish assemblages. Samples collected in both the pre- and post-impoundment periods were randomly dispersed on the SOM, and, therefore, a clear and significant pattern of separation between samples collected during these two time periods was not found. Mean and maximum water depth, which is correlated with ground water level, did not significantly separate the pre- and post-impoundment samples. However, we found significant differences between the two periods for water temperature, pH, conductivity, DO and current velocity, but abundances of fish species (summarized in the clusters identified by the SOM) did not differ significantly. Instead, the validity of the clusters distinguished by the SOM was confirmed by significant differences in some biotic variables: species richness, equitability and log transformed total abundance. Indicator species values identified the most preferred cluster (and respective complex of environmental factors) for a given species. Only one cluster did not contain any significant species indicator values, but it was dominated by samples from the Furnas Stream, which was the only effluent that could be entered by fish from the main river channel after the damming owing to its location below the dam, which has no fish ladder. Copyright # 2008 John Wiley & Sons, Ltd. key words: tropical region; impacts of reservoirs; stream fish assemblages; artificial neural network; indicator species Received 3 June 2008; Revised 11 September 2008; Accepted 15 September 2008 INTRODUCTION Tens of thousands of dams along rivers have been constructed worldwide (Petts, 1984; Nilsson and Berggren, 2000). This number will probably significantly increase in the future (Allan, 1995), and the number of studies will increase accordingly (e.g. Kinsolving and Bain, 1993; Orth and White, 1993; Koryak and Hoskin, 1994; Travnichek et al., 1995; Matthews, 1998; Penczak et al., 1998; Jackson et al., 2001; Matthews and Marsh- Matthews, 2007). Of these studies, the number of papers is highest for evaluation of changes to assemblages in main-channel tailwaters (downstream of dams), and therefore more and greater changes for these fish populations have been recorded (Poff et al., 1997; Poff and Hart, 2002). Less attention was paid to a river section located upstream from the reservoirs in main channels (Kruk and Penczak, 2003; Cumming, 2004; Matthews and Marsh- Matthews, 2007), which retained some naturalness after damming (Petts, 1984; Penczak et al., 1998, Penczak and Kruk, 2005). This is manifested in the fact that they have water temperature, discharge and velocity, which are not much altered but dependent on climate only (Petts, 1984; Penczak et al., 1998; Penczak and Kruk, 2005). Thus, the effect of the dam is limited to that of precluding riverine species migration (Petts, 1984; Kruk and Penczak, 2003). Very little is known about fish in dammed first order streams above low dams (Cumming, 2004) or in streams that are tributaries of rivers that were later dammed (Liensch et al., 2000; Matthews and Marsh-Matthews, 2007). RIVER RESEARCH AND APPLICATIONS River. Res. Applic. 25: 1013–1024 (2009) Published online 15 October 2008 in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/rra.1200 *Correspondence to: Tadeusz Penczak, Department of Ecology and Vertebrate Zoology, University ofLo ´dz ´, 12/16 Banacha Str., 90-237 Lo ´dz ´, Poland. E-mail: penczakt@biol.uni.lodz.pl Copyright # 2008 John Wiley & Sons, Ltd.