Influence of an artificial dam on the fish assemblage of a Neotropical stream

The objective of this study was to evaluate the influence of a dam on the ecological parameters of the fish assemblages of a stream of second order of the upper Paraná River basin, southern Brazil. Samplings of abiotic data and ichthyofauna occurred quarterly between September 2012 and June 2013. Were collected 3,128 individuals in six orders, eight families and 13 species. The Cluster analysis (Jaccard) based on species composition indicated the separation of the sampling units into two groups, demonstrating the dissimilarity between upstream and downstream units. The artificial dam was characterized as a discontinuity factor for diversity patterns and ecological processes related to biota. Thus, this study has indications that the creation of dams can significantly affect the richness and abundance of fish species of small lotic systems, as streams, by modifying their environments.


Introduction
The neotropical region has one of the highest diversities of freshwater fish on the planet, with estimates ranging from 4,400 (Reis, Kullander, & Ferraris Jr., 2003) to 8,000 (Schaefer, 1998) species. In Brazil, it is estimated that there are more than 2,400 species of freshwater fish (Buckup, Menezes, & Ghazzi, 2007) and approximately 600 species are present in the Paraná River basin (Agostinho et al., 2004). About 50% of the neotropical ichthyofauna comprises small species from small rivers or streams (Castro, 1999).
According to the River Continuum Concept (Vannote, Minshall, Cummins, Sedell & Cushing, 1980), in a lotic water system, such as a stream, increased complexity and habitat heterogeneity in the direction upstream to downstream causes an increase in the diversity of species and trophic structure in the same direction. However, continuity in the structure of the headwater of a stream's ecosystem can be broken, mainly by anthropogenic interference that can disrupt patterns of diversity and ecological processes related to the biota (Ward & Stanford, 1983). A common anthropogenic interference in basins in Brazil is the creation of dams. Dams cause various impacts on aquatic ecosystems, affecting physical, chemical and geomorphological characteristics. Blocking a lotic environment modifies the primary productivity and channel structure, changes the diversity of fish and invertebrates by blocking migration, and causes changes in the planktonic biomass (Agostinho, Gomes, & Pelicice, 2007).
The effects of a dam depend on its size and the surrounding ecological and environmental structure (Poff & Hart, 2002). In Brazil, there are more than 700 large reservoirs, and the area most affected by dams is the upper Paraná River basin (Agostinho, Pelicice, & Gomes, 2008). There are several works that focus on the ichthyofauna of dammed areas in neotropical river basins Gubiani, Gomes, & Agostinho, 2012;Vasconcelos, Alves, & Gomes, 2014;Ortega, Júlio, Gomes, & Agostinho, 2015); however, there are few studies that characterize the interference of dams on fish assemblages of small aquatic environments, such as streams (Gardner, Coghlan Jr., Zydlewski, & Saunders, 2011). Streams are hydrologically very important for surrounding rivers of a hydrographic basin because they directly influence the recharge (Alekseevskii, Grinevskii, Efremov, Zalavskaya, & Grigoreva, 2003), provide habitats for benthic invertebrates that process thick and fine organic particulate (Vannote et al., 1980), and act as breeding grounds for fish species that occupy other parts of the drainage basin (Rêgo, Pinese, Magalhães, & Pinese, 2008).
The persistence and conservation of fish in a lotic system, such as a stream, depends on the maintenance of appropriate conditions of the aquatic and terrestrial environments. Therefore, the availability and quality of food resources, reproduction sites, and the microhabitats for fish species in lotic systems are influenced in multiple ways by both internal and external factors (Caramaschi, Mazzoni, Bizerril, & Peres-Neto, 1999), which can be altered by human impacts, such as the creation of a dam.
Thus, the aim of this study was to evaluate the influence of a dam on the ecological parameters of the fish assemblages of Chumbeiro Stream, a second order stream in the upper Paraná River basin in southern Brazil. The hypotheses of this study are: i) the continuity in the patterns of richness and diversity of stream fish species broken by the artificial dam, ii) the environmental factors in sections located upstream and downstream of the dam influence the fish assemblage composition of Chumbeiro Stream, and iii) the abundance of the fish assemblages changes temporally, independently of the artificial barrier.

Study area
The Paranapanema River basin (22º-26º S, 47º-54º W) is in the system of the upper Paraná River basin, in southern Brazil, extends approximately 930 km, and has a drainage area of 109,600 km 2 . About 630 km of the Paranapanema River are located in São Paulo and 300 km are on the border between the states of São Paulo and Paraná (Sampaio, 1944). Chumbeiro Stream (23°11'8,47" S, 49°57'56,95" W), in the Paranapanema basin, is a second order stream in the municipality of Jacarezinho, Paraná State, Brazil. The stream is the sixth tributary of the left margin of the Água Feia River, with a drainage area of about 2 km. It starts on the margin of the highway BR-153 and runs to the Refúgio de Vida Silvestre de Jacarezinho (RVSJ) conservation unit in the municipality of Jacarezinho, Paraná State. The Fish and abiotic data were sampled at six points upstream and downstream of Chumbeiro Stream. Three points were located upstream (P1, P2 and P3) and three were located downstream (P4, P5 and P6) of a small dam constructed in order to provide water for irrigating plants in RVSJ (Figure 1).

Sampling
The abiotic data and fish were sampled quarterly between September 2012 and June 2013. The physical and chemical characteristics measured at each sample unit were the following: maximum width and depth of the channel (with the aid of a 10 m tape measure), water temperature, and pH (Tecnopon ® MPA-210P). The fish were collected with two round sieves (0.5 cm mesh x 0.78 m diameter) and a rectangular sieve (0.5 cm mesh x 1.02 m long x 0.81 m wide). Each section was blocked with 3 mm mesh networks in an area of 50 m. The sampling occurred during the day for a standard time of 50 minutes, with five collectors. The individuals captured were anesthetized, sacrificed by immersion in eugenol (active principle: phenoliceugenol 4-alil-2-metoxifenol-C10H12O2, derived from the stems, flowers and leaves of the trees Eugenia caryophyllata and Eugenia aromatica; Griffiths, 2000), fixed in 10% formalin solution and then preserved in 70% alcohol. Identification of species was carried out according to Reis et al. (2003), Oyakawa, Akama, Mautari, and Nolasco (2006) and Graça and Pavanelli (2007

Data analysis
The structure of the Ichthyofauna of Chumbeiro Stream was analyzed in terms of absolute frequency and abundance, and relative frequency and abundance, for each point, as well as by the Shannon-Wiener diversity index (H' = -Σpi logpi, where pi is the relative frequency of species i), equitability (J = H'/H' max , where H' is the Shannon diversity index and H' max [maximum diversity] is the logarithm of species richness) and Simpson's dominance index (D = Σ pi²; Ludwig & Reynolds, 1988). In addition, a cluster analysis was made with the species composition (Jaccard index) for the six sampling points. An analysis of variance (ANOVA One Way) was also made, according to the assumptions, to detect possible significant differences (p < 0,05) for the abiotic data and composition and abundance of the fish species among all sampling points separately, between sampling points upstream and downstream of the artificial dam, and among sampling periods. Finally, a canonical correspondence analysis (CCA) was made in order to relate the abiotic factors with the fish assemblage structure. The diversity and cluster analyses were performed using Statistica 7.1 (Statsoft, 2005), the analyses of variance with ECOSIM (Entsminger, 2012), and the CCA with the Vegan package (Oksanen et al., 2013) of R (R Core Team, 2013).

Results
There was a significant difference between the abiotic data of the sampling points only for the maximum depth parameter (ANOVA: F = 4.596 and p < 0.05), where P3, P5 and P6 were the points with the greatest depths ( Figure 2). Maximum width, pH and temperature were not significantly different among the sampled points.
In total, 3,128 specimens were collected, including six orders, eight families and 13 species (Table 1). Chacaciformes was the most abundant order (80.53%) and had the highest species richness (46%), followed by Cyprinodontiformes (relative abundance: 9.37%; relative richness: 15.39%) and Siluriformes (relative abundance: 0.67%; relative richness: 15.39%). Perciformes, Gymnotiformes and Synbranchiformes accounted for only one species each. Astyanax bockmanni, Astyanax sp., Bryconamericus iheringii and Hypostomus ancistroides were the only species shared between sampling points upstream and downstream of the artificial dam. The other species were found only at points downstream of the dam.  Astyanax bockmanni had the highest (55.85 %) and Coridoras aeneus had the lowest (0.03 %) relative abundance. Regarding the diversity parameters, the mouth of Chumbeiro Stream (P6) had the highest value in the Shannon-Wiener diversity index and the lowest dominance. The least diverse unit was P1, and P5 showed the least equitability. The results of the diversity index are presented in Table 2. The cluster analysis (Jaccard index) indicated the separation of two sampling units in the species composition, demonstrating the similarity between sampling units upstream of the artificial dam (P1, P2 and P3), and among P4, P5 and P6 located downstream of the dam. The sampling units P1 and P3 presented the same species composition (Figure 3). The analyses related to the temporal pattern and assemblage abundance of Chumbeiro Stream did not show significant differences for the four sampling periods. However, it can be seen that the highest abundance was recorded during the second sampling period (December 2012) and the lowest was recorded during the fourth period (June 2013).
The CCA diagram did not demonstrate a significant relationship of the width, depth and temperature with the fish assemblages (ANOVA: F = 0.54 and p > 0.05). However, abiotic factors had a greater influence on the fish assemblage of P5 (Figure 4).

Discussion
The higher species richness and abundance of the order Characiformes found in this study reflect a common standard for streams of the upper Paraná River (Abes & Agostinho 2001;Castro et al., 2003;Araújo, Delariva, Bonato, & Silva, 2011) and for the Neotropics (Lowe-Mcconnell, 1999). The high representativeness of order Cyprinodontiformes is due to the large abundance of Poecilia reticulata captured in the sections downstream of the artificial dam (P4, P5 and P6). Poecilia reticulata, known as lebiste, is not native from the upper Paraná River system; it was introduced in order to control the malaria mosquito (Langeani et al., 2007) and was released from aquariums (Oliveira et al., 2014). This species is a cause for concern because it is considered invasive and could possibly harm the biodiversity of ecosystems (Mack et al., 2000). Gymnotus inaequilabiatus was another species found in this study that is not native from the upper Paraná River system; however, its origin in this basin is unknown (Langeani et al., 2007).
Concerning the temporal analysis, the largest abundances found in the summer (December 2012) may reflect the increase in fish metabolisms at the end of the winter (September 2012), which causes an increase in reproduction after the end of this drier and colder season (Oliveira, Goulart, & Minte-Vera, 2004). Thus, for most species, an increase was found in the number of individuals during the hottest and rainiest season (summer -December 2012) and a decrease in June 2013 (early winter).
The Shannon-Wiener diversity index values among the sampling units support the River Continuum Concept by Vannote et al. (1980). The artificial dam situated downstream of P3 is a factor that may have accentuated this trend, preventing the flow of the downstream populations to the upstream units and reducing the diversity in them.
As for equitability, two sampling units upstream of the artificial dam (P1 and P3), which showed higher values in the index,also had the smallest richness values, with only three species each (Astyanax bockmanni, Astyanax sp. and Bryconamericus iheringii). The low species richness in P1 may have been caused by lower values of depth and flow rate, which reduces the possibility of new habitats, microhabitats, and reproduction sites (Kaufmann & Pinheiro, 2009). However, the dam can be considered responsible for the lower richness upstream, also verified in P3, because these units does not receive species from the Água Feia River, where the Chumbeiro Stream ends. Thus, the artificial dam has disconnected the Água Feia River from the upstream units of Chumbeiro Stream, and the dam (in P3) probably made the upstream units more similar to each other and dissimilar to the downstream units. This makes the artificial dam a discontinuity factor in the patterns of diversity and ecological processes related to the biota. Silva, Guibiani, Piana, and Delariva (2016) detected similar results with a natural barrier, checking greatest richness of unique species from the downstream of the barrier, and found a significant difference in the richness, equitability and abundance between upstream and downstream. Also in a survey conducted in China (Sui, Lu, Yan, Chen, & Jia, 2014), in streams of the first to the fifth order, the construction of dams apparently decreased the number of species upstream substantially.
The dam affects the stream fish communities in various ways, but two of the main direct impact mechanisms are the alteration of habitat upstream and downstream of dam and the fragmentation in the flow of fish populations. Consequently, this fragmentation of the fish populations can cause a reduction in gene flow, resulting in lower effective size of the populations and deleterious effects of endogamy (Hayes, Dodd, & Lessard, 2006). Transposition stairs of fishes can be built, but it not solves the problems caused by barrier, since the fish cannot cross this system efficiently (Fontes Jr., Castro-Santos, Makrakis, Gomes, & Latini, 2012). Fish cannot leave the stream or river that they are living in or migrating through to bypass a barrier, unlike other animals (Ottburg & Blank, 2015). Thus, the dam removal can restore connectivity of the river for fish movement (Poulos et al., 2014).
The sampling units P1 and P3 were without representative species in the CCA, were more Acta Scientiarum. Biological Sciences Maringá, v. 38, n. 3, p. 305-311, July-Sept., 2016 equitable and had lower species richness. The units P1 and P3 were also without representative species in the CCA, were more equitable and presented lower species richness, which was different from P5, which had several associated species. This result was expected because P5 was the only section with exclusive species (Corydoras aeneus, Phalloceros harpagos and Synbranchus marmoratus), in addition to being the unit with the highest richness and Shannon-Wiener diversity values, and the larger number of fish species increased the possible associations.
The results of this study indicate that Chumbeiro Stream, of the upper Paraná River, is strongly influenced by the environmental disruption caused by human impact, which is the artificial dam. Thus, this research indicates that the creation of dams, even when small, can significantly affect the richness and abundance of fish species from small lotic systems, such as streams, by modifying their environments.