Assessment of fish assemblages in streams of different orders in the Upper Paraná River basin , Central Brazil

The aim of this study was to test whether the richness observed and the biomass per trophic group of fish assemblages vary depending on the order (1st and 2nd) of the streams located in three different basins of the Upper Paraná River Basin, Central Brazil. Samples were collected between April and September, 2009, in 27 streams of the Meia Ponte, Piracanjuba and Santa Maria River basins. A total of 4,879 specimens were collected distributed in 59 species and 19 families. The statistical analyses carried out indicate that the observed richness and biomass of omnivore fish were influenced by the interaction of two factors: stream order and basin. The 2nd order streams located in the Santa Maria basin presented significant differences in the observed richness and omnivore biomass when compared to i) 1st order streams in the same basin (only richness) or in the Piracanjuba and Meia Ponte basin; ii) 2nd order streams in the Piracanjuba (only omnivore biomass) and Meia Ponte Rivers basins. Results are discussed considering the influence of geomorphic processes on fish assemblages and food availability.

Identifying patterns of distribution of richness and abundance of species and relating these to environmental variations is one of the major goals of studies on the ecology of streams (Giller & MalMqvist, 1998).One way to explain these patterns of richness and abundance is by using the physical parameters of the drainage network (Jones iii et al., 1999;taylor et al., 2006;Fialho et al., 2007;arauJo & teJerina-Garro, 2009;KashiwaGi & Miranda, 2009), such as the drainage basin and stream order.
A drainage basin is a set of small water bodies which come together at a certain point to form a larger body of water.It is considered an isolated unit, as it is bordered by oceans and/or large areas of land, which act as barriers to the dispersal of species, thereby leading to differentiation between aquatic communities located in different drainage basins and, in some cases, to speciation (huGueny et al., 2010).On the other hand, drainage basins can differ from one to another with respect to climate, vegetation, geology and topography which control the geomorphic process and influences on aquatic ecosystems (MontGoMery, 1999), thereby bringing differences between fish assemblages in different drainage basins (MaGalhães et al., 2002).
Stream order (strahler, 1957) is useful for evaluating fish distribution in streams because of its influence on fish assemblage, but by itself, it is not a pervasive organizer of lotic fish assemblages (Matthews, 1986).Studies about this influence indicate that increasing stream order results in the increased richness (whiteside & Mcnatt, 1972;Platts, 1979;osborne & wiley, 1992;ibañez et al., 2009), abundance (sMith & KraFt, 2005) and diversity (harrel et al., 1967;GorMan & Karr, 1978) of fish assemblage in temperate and tropical streams.This trend is attributed to i) addition and/or replacement of species (beecher et al., 1988) brought about by changes in the abiotic characteristics of stream (sMith & KraFt, 2005), which tend to increase the complexity of the aquatic habitat represented, for example, by stream depth and width, bottom type and current (GorMan & Karr, 1978;Platts, 1979); ii) shorter distance to downstream source population; iii) reduced barriers to migration from downstream locations; iv) more stable habitats downstream (wineMiller et al., 2008).
In headwater courses (1 st to 3 rd order) the main source of energy available is allochthonous organic matter (vannote et al., 1980;uieda & Motta, 2007), which is used as a direct (redFord & Fonseca, 1996) or indirect (walKer et al., 1990) source of food by local fish assemblages.The trophic composition of these assemblages is persistent over time (MeFFe & berra, 1988) and is composed of fish which occupy almost the entire spectrum of trophic niches which occur in aquatic communities (wineMiller et al., 2008).Insectivores dominate headwater courses (Paller, 1994), whereas omnivores tend to increase with stream order (sMiley Jr et al., 2005;wineMiller et al., 2008;ibañez et al., 2009).
Against this background, the aim of this study was to test whether (i) the richness observed and (ii) the biomass per trophic group of fish assemblages vary depending on the order (1 st and 2 nd ) of the streams located in three different basins of the Upper Paraná River Basin, Central Brazil.
The climate of the region in which the sampled basins are situated ranges from humid to sub-humid, according to the Köppen classification, with two distinct seasons, the rainy season from October to March and the dry season from April to September (brasil, 1977).Fishes were sampled during the dry season when flows are low thus making for a more efficient capture (Pease et al., 2012).
The sections sampled in each stream were selected according to their accessibility.Two 100-meter stretches were demarcated and geo-referenced in each stream, one downstream and the other upstream from the access point, 15 m distant from each other.For all analyses, data from the two stretches were grouped.All streams sampled were located away from urban areas and surrounded by grasslands, except for the P17 stretch, which was encircled by a sugarcane plantation.All stretches presented riparian forest which, in some cases, had been replaced by grass for feeding cattle (P5) or swamps (P9).
Fish sampling was conducted in the morning (7:00-12:00) using electrofishing and following the modified protocol of Mazzoni et al. (2000).Each 100 m stretch was covered just once instead of three times in a downstreamupstream direction.The sampling effort was 1 h/100 m which was repeated in each stream sampled.
Afterwards, the fish were packed in plastic bags, containing an identification tag, fixed in 10% formalin and subsequently preserved in 70% alcohol.In the laboratory, the fish were weighed (g), measured (standard lengthmm) and identified to the lowest taxonomic level possible.Classification of the species by trophic guild was based on the literature available, with preference given wherever possible to studies undertaken in lotic environments of the Upper Paraná River basin (Tab.II).In the case of species identified at genus level, trophic guild of species of the same genus was used.Thus, the following guilds were considered: (i) Carnivores -species which feed on fish and/or molluscs, decapods, microcrustaceans and other invertebrates but not insects; (ii) Detritivores -fish which feed on detritus (sediments and periphyton); (iii) Insectivores, which consume aquatic or terrestrial insects (adult or larvae) and arachnids; (iv) Omnivores -species which feed indiscriminately on vegetal and animal.
The difference between the fish assemblages was evaluated by using two generalized linear mixed models (GLMM) with nested design separately.This was chosen since the sampled streams are 1 st and 2 nd order and are grouped (nested) within each of the three basins considered (Tab.I).The first analysis was performed considering observed richness as a dependent variable, stream order as a fixed effect and basin and abundance as a random effect; the second analysis used data per trophic guild (biomass) as a dependent variable, stream order as a fixed effect and basin as a random effect.The use of biomass is appropriate when dealing with trophic guilds because it represents all or part of the population supported by the same energy source (burns, 1989).Both analyzes were followed by a post-hoc Tukey analysis.All tests were carried out in the Statistica 8.0 software (statsoFt inc., 2007).

RESULTS
A total of 4,870 individuals were collected distributed in 59 species and 19 families (Tab.II).The GLMM with nested design indicates that the observed richness and omnivore biomass of the fish assemblages sampled were influenced by the interaction of the stream order and basin factors.Richness is not influenced by abundance (Tab.III).
The 2 nd order streams located in Santa Maria basin presented fish assemblages with the highest values of observed richness (mean = 32 species; Fig. 2) and omnivore biomass (mean = 487 g; Fig. 3).These values are significantly different from the fish assemblages of 1 st order streams in the same basin (only richness), streams of the Piracanjuba River basin (1 st order for richness; both stream orders for omnivore biomass; Tab.IV) or those of the Meia Ponte River basin (both stream orders for richness and omnivore biomass; Tab.IV).

DISCUSSION
Only the interaction of the basin and stream order shows an influence on fish assemblage.This result expresses the influence of geomorphic processes on fish assemblages (MontGoMery, 1999;MaGalhães et al., 2002), that is, the Santa Maria River basin is characterized by the widest and deepest stream channel with greater water velocity than streams of the other basins.These characteristics increase habitat heterogeneity (GorMan & Karr, 1978) and availability of resources, thereby increasing the potential number of niches, which in turn facilitates the coexistence of more species (wineMiller et al., 2008;huGueny et al., 2010).This seem be expressed in this study by the differences among streams of 2 nd order of the Santa Maria River basin and streams of 1 st and 2 nd order of the Piracanjuba and Meia Ponte Rivers basins.On the other hand, the absence of differences between the streams of the Piracanjuba and Meia Ponte Rivers basins can be related to the similarity of the habitat, supporting the idea that streams of equal order, even if located in different basins, would present similar fish assemblage in terms of species richness and abundance (vannote et al., 1980).
The results of this study also suggest that the structure of fish assemblages increases in richness and omnivore biomass in accordance with increases in stream order (from 1 st to 2 nd order).Different studies report this trend for fish richness [e.g., Platts (1979) from 1 st to 5 th order; sMith & KraFt (2005) -1 st to 3 rd ; osborne & wiley (1992) -1 st to 5 th ].In this study, this trend seems to be supported by changes in stream characteristics (width, depth and current increase as stream order increases) as  (2009) in tropical and temperate streams.However, it is necessary to stress that other factors not measured in this study can also influence these results such as the shorter distance of the fish assemblage of the streams of 2 nd order from downstream source fish populations, the reduced barriers to fish migration from downstream locations to 2 nd stream order, and more stable habitats of 2 nd stream order than those of 1 st order (wineMiller et al., 2008).Additionally, the extent of anthropogenic activities in each basin can influence on obtained results, once this induces changes of the fish assemblage structure as observed by Fialho et al. (2008) in the Meia Ponte basin.
The increase of omnivore biomass as stream order increases can be a consequence of food availability.The supply of food resources in a stream of 2 nd order comes from two main sources, the adjacent riparian forest and the resources brought by its tributaries.In other words, food availability tends to be greater than that of streams located at the headwaters, 1 st order in this case (redFord & Fonseca, 1996;raKocinsKi et al., 1997).However, even in streams of 2 nd order food availability is not constant along time and is influenced by environmental constraints, which led to trophic constraints of fish assemblages and finally to differing proportions of trophic guilds (ibañez et al., 2009), omnivores in this case.In tropical streams one environmental constraint influencing food availability and trophic composition is the climate characterized by rainy and dry seasons, as observed in the basins sampled.Accordingly to wineMiller et al. (2008) during the dry season food resources are limited and in these circumstances some tropical fish species present more specialized feeding habits based on their morphology that allows for relative foraging efficiency.In this situation, the omnivore species could be the least influenced due to their broad dietary options (uieda & Motta, 2007).
It is concluded that the fish assemblage is influenced by the interaction of stream order and basin factors resulting in a significant increase of richness and omnivore biomass from streams of 1 st to 2 nd order.However, additional studies including higher orders than those considered are necessary aiming to verify the achievement of results obtained.
Fig. 1.Location of streams sampled (black circles) from April to September, 2009, in the Upper Paraná River basin, Central Brazil.The black squares represent the main urban areas.The black area represents the reservoir at the Itumbiara hydroelectric plant.

Figs
Figs 2, 3. Plot resulting of the generalized linear mixed models (GLMM) with nested design analysis for observed fish richness (2) and omnivore trophic guild (3) per stream order and basin sampled in the Upper Paraná River basin, Central Brazil, between April and September, 2009.The vertical bars represent the confidence interval of 95%.
Iheringia, Série Zoologia, Porto Alegre, 104(2):175-183, 30 de junho de 2014Tab.II.Number of individuals and trophic guild according to the literature per fish species collected in the 27 streams sampled in the Upper Paraná River basin, Central Brazil, between April and September, 2009.Tab.III.Statistics of the generalized linear mixed models (GLMM) with nested design (stream order and basin) analysis for observed fish richness and biomass per trophic guild of the 27 streams sampled in the Upper Paraná River basin, Central Brazil, between April and September, 2009 (*, significant differences, P<0.05; DF, degree of freedom).