1.Introduction

Hydrologic connectivity of riverine systems is crucially important to fish and other aquatic organisms. Nonetheless, these ecosystems are impaired by impoundments with different purposes (e.g. hydropower, water supply and flood control). Loss of connectivity is frequently linked to extinction of entire populations of freshwater fishes (^{Sheer and Steel, 2006}; ^{Olden et al., 2010}), changes in hydrologic regimes (^{Graf, 1999}), introduction of non-native piscivorous (^{Holmquist et al., 1998}; ^{Pelicice and Agostinho, 2009}) and blocking of migratory routes (^{Roscoe and Hinch, 2010}), thus contributing to reduce the diversity and change the assemblage organisation.

Although the problems associated with impoundment have been addressed in a variety of reviews and scientific articles (^{Baxter, 1977}; ^{Cumming, 2004}; Poff et al., 2007), until now, the mitigation measures have been limited to fish passage facilities. Several types of fish passages (e.g., fish ladders, lifts, locks and trap-and-truck operations and “nature-like” fish ways; reviewed in ^{Roscoe and Hinch, 2010}) have been used to minimise the negative effects of the river fragmentation on fish assemblages. However, some mechanisms have been criticised for the low efficiency in facilitating two-way movements, high selectivity and inefficiency in allowing downriver return by adults, eggs, and larvae (^{Agostinho et al., 2007}; Godinho and Kynard, 2009, ^{Pompeu et al., 2011}).

In Brazil, there is a crescent investment on clean energy provided by hydropower, mostly because of its great water resources. Only in the main channel of Paraíba do Sul River there are four large dams, three low head dams (head height < 15 m), which vary in size, shape, function and operational routine. Although it is one of the most important Brazilian rivers, draining developed states in the country (São Paulo, Rio de Janeiro and Minas Gerais), little is known about the consequences of its regulation by dams (^{Bizerril, 1999}; ^{Polaz et al., 2011}). Every dam has unique characteristics and, consequently, the scale and nature of the environmental changes are highly site-specific (McCartney, 2009). The traditional dams built for hydroelectric purposes blocks the entire river course, causing fragmentation of the river and storing a large volume of water. On the other hand, low-head dams are considered to have less environmental impact because maintain the fluvial connectivity, operating as “run-of-the-river” mode, not generating large reservoirs.

The effects of traditional dams i.e., those built for hydroelectric purposes, which block the entire river course and store a large volume of water, have been documented by a number of studies (^{Rosenberg et al., 2000}; ^{Poff and Hart, 2002}; ^{Agostinho et al., 2008}; ^{Pompeu et al., 2011}). On the other hand, less research has been undertaken on the impacts of non-traditional dams. These particularities allow us to make investigations focusing on small scale and determine whether the connectivity between zones in a low-head dam (Santa Cecília Dam, Paraíba do Sul River) that have a side channel connecting permanently the reservoir and the downriver zone (partial blockage) is sufficient to ensure the maintenance of ichthyofauna structure. Unlike the traditional fish ways that facilitate upriver migrations, here we examine a fish passage which probably allows only downriver movements. We tested the hypothesis that similarity of the fish assemblages differ between these two zones (reservoir vs. downriver zone). The aim was to evaluate possible effects of this dam on fish fauna organisation, considering the spatial and temporal scales.

2.Materials and Methods

3.Results

A total of 1687 individuals comprising 27 species were collected in the downriver zone, while the reservoir had 879 individuals and 23 species (Table 1). Only four fish species occurred in more than one-third of the samples: Astyanax bimaculatus (Linnaeus, 1758), Oligosarcus hepsetus (Curvier, 1829), Rhamdia quelen (Quoy and Gaimard, 1824) and Hypostomus affinis (Steindachner, 1877) with the former being present in 59.4% of all samples. A wide size range of the fish population was sampled, ranging from the smallest A. bimaculatus with 6.5 cm Total Length to the largest Hoplias malabaricus (Bloch, 1794) with 42.5 cm Total Length. Six species were recorded exclusively from the downriver zone, while only two species were exclusive from the reservoir. All the seven migratory species were recorded in the downriver section, but only five in the reservoir. Moreover, two non-native species were recorded in both zones and one more non-native species was exclusive of each zone.

Abiotic environmental values of the different sampling zones/seasons are summarised in Table 2. The temperature was comparatively higher during wet season irrespective of zone. Both zones had higher dissolved oxygen, conductivity and transparency in the dry season.

Principal component analysis on the abiotic data yielded two axes with eigenvalues greater than one, explaining 71.6% of the total variance (Table 3). Abiotic variables which most contributed to PC1 were temperature (positively associated to this axis), conductivity, pH and transparency (negative related). The most variation in the abiotic data (PC1 = 53.28%) was explained by seasonal patterns, discriminating wet and dry season samples (Figure 2). Dissolved oxygen was inversely associated to PC2 that explained only 18.3% of the total variance.

MDS analysis suggests that the spatial component is more relevant to determine the fish assemblage organisation than the seasonality, i.e. distinct fish assemblages were found for each zone (Figure 3). Moreover, linear regression of fish assemblage (MDS1) against abiotic variables (PC1) was no significant (r² = 0.08; p = 0.307).

Thirteen species had significant indicator values according to the indicator species analysis (Table 4). The downriver zone had the greatest number of indicators species, among them three were rheophilics (Leporinus copelandii, Prochilodus lineatus and Pimelodus fur), with seven being indicators for the wet season and only one for the dry season. Hoplias malabaricus, Geophagus brasiliensis (Quoy & Gaimard, 1824) and Rhamdia quelen were indicators for the reservoir zone during the dry season, while Hoplosternum littorale (Hancock, 1828) was indicator for the reservoir zone during the wet season.

4.Discussion

A major change in the ichthyofauna organisation was detected in the spatial scale, i.e. the impoundment impacts on fish distribution, while seasonal variation between wet and dry seasons played a pivotal role in abiotic environmental patterns. Although some fish species were typical from wet or dry season, their distribution are probably more related to biological rhythm, foraging mode, and/or behavioural adaptations, than with limnological variability.

Abiotic variables had a strong influence in the seasonal patterns, explained by variation in temperature, transparency, pH and conductivity. High temperature values were directly related with the warm summer season, which coincides with rainy periods in rivers from Southeastern Brazil. Low water transparency found during the wet season can be regarded a consequence of the increased sedimentary inputs from erosion, which is characteristic of the Paraíba do Sul River during flood periods (^{Pfeiffer et al., 1986}). Water conductivities tend to be lower during wet season (^{Winemiller and Jepsen, 1998}), and influenced byanthropogenic activities because of inputs of organic matter, and its later decomposition by heterotrophic organisms (^{Matthews, 1998}; ^{Guarino et al., 2005}). The importance of abiotic factors in structuring fish assemblages is thoroughly recognised (^{Schlosser, 1982}; ^{Jackson et al., 2001}; ^{Oberdorff et al., 2001}). However, in this study, similarly to the findings of ^{Gubiani et al. (2010)}, the variation detected in abiotic variables was not very meaningful to define fish composition.

A more abundant and richer assemblage was found in the downriver compared with the reservoir zone, which is a widely documented consequence of dam construction (e.g., ^{Gomes and Miranda, 2001}; ^{Oliveira et al., 2004}; ^{Santos et al., 2010}). According to ^{Williams et al. (1998)}, it is explained by large-scale changes in temperature, turbidity, flow, allochthonous nutrient inputs, and availability of food resources regimes. Considering the absence of relationships between fish assemblage and selected environmental variables, we suggest that regional changes caused by the impoundment is a main factor responsible for spatial fish distribution; increasing the aggregation of fishes below the dam due to the obstacle for upriver movements, and preventing the colonisation of species above the dam.

River impoundments generally exert negative effects on native fish species, especially on migratory species, disrupting longitudinal connectivity and obstructing migrations routes (^{Joy and Death, 2001}; ^{Fukushima et al., 2007}). Furthermore, the accumulation of migratory fishes immediately below dams is well documented (^{Taylor et al., 2001}; ^{Gehrke et al., 2002}) and were found in this study, since several species associated to downriver zone are migratory or rheophilic species - e.g., Leporinus copelandii Steindachner, 1875; Pimelodus fur (Lütken, 1874); and Prochilodus lineatus (Valenciennes, 1837). In addition, all these species were associated with wet season, probably due to reproductive purposes. It is widely reported that tropical migratory fishes need environmental stimuli (rain and floods) to trigger migrations and spawning (^{Lowe-McConnell, 1975}; Vazzoler et al., 1997).

Prochilodus lineatus distribution is heavily impacted by successive impoundments in Paraíba do Sul River, which must be preventing reproductive migrations. Here, only two individuals were caught in the reservoir zone. It is a long-distance migratory species that depend on the flood pulse to complete the life cycle (^{Gubiani et al., 2007}). Early experimental studies revealed that Prochilodus species can migrate between 450 and 600 km y^–1 (^{Bayley, 1973}; ^{Agostinho et al., 1993}). According to ^{Antonio et al. (2007)}, it is able to make two-way migrations, and tend to avoid the impounded regions in their reproductive period.

We found a decrease in abundance, especially largesized migratory species in the reservoir zone, which are replaced by small-sized, generalist (in terms of habitat and food requirements) and opportunist fishes (^{Agostinho et al., 2007}). Although the Paraíba do Sul River has lesser large-sized species compared with other larger Brazilian River Basins (Araújo et al., 1996), numbers of large-sized species could be underestimated because of the mesh size of nets that were at most 75 mm stretched mesh. Among the species inhabiting the reservoir zone, the following stand out as well suited to reservoir environment: G. brasiliensis which is a highly adapted to lentic environments (^{Bizerril, 1999}) and has generalist feeding habits with great trophic plasticity to support changes in food resources availability (Meschiantti, 1995; Aguiaro and Caramashi, 1998); and H. littorale that is a typical fish of muddy habitats with physiological adaptations to breathe atmospheric oxygen and, consequently, to tolerate impacted environments (^{Chagas and Boccardo, 2006}).

Another factor that contributes to decrease and/or displacement of native species is the predation pressure exerted by carnivorous fishes. Usually, in reservoirs, this guild is composed by non-native species, such as Cichla kelberi Kullander & Ferreira, 2006 and Plagioscion squamosissimus (Heckel, 1840) (^{Bennemann et al., 2006}; ^{Pelicice and Agostinho, 2009}). Here, curiously, this group was dominated by two native species, H. malabaricus and R. quelen. The persistence of H. malabaricus in the reservoir zone is probably favoured by its prey-ambush foraging behaviour and strong territoriality during the reproductive season (^{Paiva, 1974}). In turn, R. quelen is an omnivorous opportunist species, eating insects, crustaceans, plant debris and small-sized fishes (^{Casatti and Castro, 2006}; ^{Oyakawa et al., 2006}), and has nocturnal habits with preference by lentic and deep waters (^{Gomes et al., 2000}). Therefore, both species showed adaptations to lead to new environmental conditions found in impounded areas.

Non-native fish predators, generally, have a higher abundance above than below large dams in tropical streams (^{Holmquist et al., 1998}). Such a distribution pattern is well established for P. squamosissimus (see ^{Oliveira et al., 2003}; ^{Agostinho et al., 2007}; ^{Terra et al., 2010}). It is a top predator fish native from Amazon River basin responsible for the decrease of native fish population in many Brazilian reservoirs (^{Bennemann et al., 2006}; ^{Pelicice and Agostinho, 2009}). In the studied system P. squamosissimus was an indicator species of downriver zone, contradicting known patterns of distribution. We believe that this bias can be related to the lateral connection between zones. According to ^{Agostinho et al. (1999)}, this species produces small, pelagic and buoyant eggs spawned in several batches during the reproductive season, and use this strategy to colonise reservoirs. However, due to the run of the river characteristics of the studied system, this strategy should be ineffective in Santa Cecília Reservoir, where eggs and larvae drift downriver while they develop. Other factors, such as largest availability of preys and the better oxygenation of the water could explain the highest abundance of this species in downriver section.

In fact, this study illustrates the negative impact of river impoundment on fish assemblages, with fish assemblage organisation significantly different between reservoir and the downriver zone. Almost all changes detect in this non-traditional dam were similar those found in traditional ones, except for the P. squamosissimus distribution, the lack of non-native piscivorous in reservoir zone, and the independence between limnological and biotic variables. Most affected species were the rheophilic and migratory ones, such as Pimelodus maculatus La Cèpede, 1803, that is much more abundant in the downriver zone compared with the reservoir. Moreover, Leporinus copelandii, Pimelodus fur and Prochilodus lineatus are probably not able to perform their reproductive upriver migration due to dam blockage. Furthermore, our findings are in agreement with ^{Matthews (1998)} who pointed out that only those species with adaptations that fit the available habitats will successfully colonise a reservoir. We also found some evidences that the lateral hydrological connection has low ecological efficiency e.g., the absence of P. fur, as well as, the limited distribution of P. lineatus in the reservoir zone. In fact, strictly rheophylic species will probably never adapt in the reservoir habitat, even if it passes to the channel. Therefore, despite lowhead dams being considered to have less impact on ichthyofauna, these dams produce semi-lentic environments with fish assemblage organisation significantly different from the fluvial system. In conclusion, we encourage biotelemetry or radio-acoustic count experiments to elucidate whether the lateral connection is able to promote fish transpositions, and to investigate the role of non-traditional dams in fish assemblage distribution.