The fish community as an indicator of biotic integrity of the streams in the Sinos River basin , Brazil

The basin of the Sinos River, located in the northeastern part of Rio Grande do Sul state, Brazil, has been highly impacted by industrial and urban activities. Water quality is low because of domestic and industrial sewage discharges. Most of the tributaries have suffered drastic structural interventions like canalisations and the removal of riparian vegetation. The aims of this study were to: 1) assess the diversity of fish at 34 sampling sites in twenty-four tributaries of the Sinos River basin; 2) quantify impact level by the Shannon-Wiener diversity index and an adapted Index of Biotic Integrity (IBI); and 3) check the interference of environmental impacts, formerly quantified in a Stream Corridor Assessment Survey (SCAS), on the fish assembly and 4) compare the relationship between the IBI with stream order. Fish sampling was performed by electric fishing in the period from April 2004 to August 2006. A total of 4,869 individuals were sampled, representing 61 species, 14 families and six orders. Significant relationships of the Shannon-Wiener index and IBI with SCAS scores and stream orders were found. Of all impacts that make up the SCAS score, only channel modifications were significantly correlated with IBI. These results indicate that the adaptation of the IBI was effective and performed better than the Shannon-Wiener diversity index when related directly to specific impact categories. The application of the IBI with the SCAS and the other variables was efficient in the tributaries of the Sinos River basin because it showed the biotic degradation in accordance with changes in physical habitat.

disturbances that degrade the Sinos River tributaries are pollution originating from domestic and industrial sewage in urban areas, eutrophication, erosion and elimination of the riparian buffer strips in agricultural areas.Discharges of toxic industrial sewage occur sporadically, causing fish kills in the river main stem.During the most severe registered fish kill, more than a 100 t died in October 2006 (FEPAM, 2007).The withdrawal of water for irrigation of rice paddies and drinking water, aquaculture, garbage deposition, and modifications of the physical characteristics of water courses like canalisations are other types of impacts that occur in these streams.
Although the composition of the fish community in the river main stem of the Sinos is well known (Petry and Schulz, 2001), the same is not true for the tributaries.The principal goals of the present study were 1) to assess the diversity of fish at 34 sampling sites in twenty-four tributaries of the Sinos River basin; 2) to quantify impact level by the Shannon-Wiener diversity index and an adapted Index of Biotic Integrity (IBI); 3) to check the interference of environmental impacts, formerly quantified in a Stream Corridor Assessment Survey (SCAS), on the fish assembly and 4) to compare the relationship between the IBI with the order of the streams.

Study area
The hydrographic basin of Sinos River is located in the northeastern region of Rio Grande do Sul state, Brazil, with its headwaters in Serra Geral and the river mouth in Canoas near the capital Porto Alegre (Figure 1).The river main stem is 190 km long and drains an area of approximately 3.820 km 2 , relative to 1.5% of the total area of the state (FEPAM, 1999).
The Sinos basin belongs to the phytogeographic region classified as Semidecidual Seasonal Forest (Teixeira et al., 1986), which today only exists on the slopes of the Serra Geral.The climate is subtropical with four well-defined seasons.
The use of water resources is related to soil occupation and to the development of the region.The upper region of the basin, originally covered by forest, is now characterised by plantations of vegetables, sugarcane and rice paddies.In the middle section of the Sinos River basin, large areas are used for rice cultivation and cattle ranching.In the lower region, densely urbanised areas with a high concentration of industries are dominant.Impacts of this area are characterised by water withdrawal for domestic and industrial use, by domestic and industrial sewage as well as huge amounts of domestic garbage (FEPAM, 1999).Only 10% of domestic sewage is treated.Industrial sewage, although more controlled by environmental defense agencies, occasionally cause severe fish kills, like the one in October 2006, when more than 100 t of fish died.

Introduction
Aquatic ecosystems have been modified in a significant way due to impacts from human activities.These activities lead to a decrease in water quality and in the diversity of biological communities, due to changing physical and chemical variables (Goulart and Callisto, 2003;Sabater et al., 2000;Tejerina-Garro et al., 2005).
Based on the use of biotic resources as indicators of ecosystem quality, Karr (1981) developed the Index of Biotic Integrity (IBI).The IBI was designed to evaluate the integrity of the fish community of small streams from the temperate climate in the Midwest region of United States.Later it was adapted to the local conditions of many other countries, as for example France (Oberdorff and Porcher, 1994), Lithuania (Kesminas and Virbickas, 2000), Africa (Hocutt et al., 1994;Toham and Teugel,s 1999), New Zealand (Joy and Death, 2004) and Argentina (Hued and Bistoni, 2005).In Brazil, most IBI studies have been carried out in the states of São Paulo and Rio de Janeiro.Araújo (1998), Araújo et al. (2003), Pinto and Araújo (2007), Pinto et al. (2006) developed studies to adapt the index to the Paraíba do Sul river in the state of Rio de Janeiro.Ferreira and Casatti (2006) did the same in the state of São Paulo, and Casatti et al. (2009) in the Upper Paraná River basin.Bozzetti and Schulz (2004) adapted IBI for streams from subtropical climates in the southern region of Brazil.It takes into account differences in the distribution, abundance, and health of the fishes, which are factors linked with the level and type of modification of the water body (Bozzetti and Schulz, 2004).Most IBIs include abundance metrics from faunal elements specific to the area of investigation.The use of these specific metrics impedes the application of a common IBI in different regions of Brazil and decreases the comparability of the results.Additionally, most studies do no quantify environmental impacts as predictor variables for IBI.The specific causes for increase or decrease of IBI scores in different watercourses therefore remain unclear.
Another approach to measure impact levels in aquatic ecosystems is the qualitative and quantitative assessment of physical and chemical impacts (Goulart and Callisto, 2003).Yetman (2001) designed an assessment of environmental impacts on streams based on the Stream Corridor Assessment Survey (SCAS) methodology for the Chesapeake basin in the state of Maryland, USA.This study did not investigate the effects of structural modifications on the fish community.Both methods, IBI and SCAS, are complementary and may be applied together, using SCAS scores as predictors for IBI.
In Rio Grande do Sul, the Sinos River is an example of a heavily impacted, multiple use watercourse, which provides drinking water for 1.6 million inhabitants (Petry and Schulz, 2006).Severe environmental modifications occur in the medium and lower parts of the basin.In these regions, the streams that compose the hydrological network of the basin pass through urban centers with high populations and industrial density (FEPAM, 1999).The main

Adaptation of IBI
The IBI suggested by Pinto and Araújo (2007) for Paraíba do Sul River, Rio de Janeiro state, Brazil, was modified for the tributaries of the Sinos River basin.This version of the IBI was more appropriate for lowland streams than the index by Bozetti and Schulz (2004) which was developed for a headwater specific fish fauna.Adaptation of metrics was necessary due to differences in the composition of fish communities found in the two environments (Table 1).
The number of native species was substituted by the total richness, since exotic species were rare (0.2% of the total number of individuals).The number of characiform species was substituted by number of individuals in the water column.This alteration allowed the consideration of individuals from other orders of the same habitat use.The number of siluriform species was substituted by the number of benthic species, since the siluriform group comprises species able to breathe atmospheric oxygen and which may occur in high abundances in organically polluted streams (Bozzetti and Schulz, 2004).Therefore, this metric could enhance the index and would exclude benthic individuals from other orders such as Characiformes of the genus Characidium.The metrics suggested by Pinto and Araújo (2007), such as the percentage of Cyprinodontiformes individuals, of omnivores, of carnivores, and the number of dominant species, which represent 90% of the total The main tributaries of this hydrographic basin are the fifth order rivers Paranhana and Rolante, and of the fourth order river of Ilha.The total extension of the hydrological network is about 4000 km (Comitesinos, 2008).Wetlands still occur in the lower parts of the drainage basin.The tributaries in the upper region of the basin are characterised by low water temperature, high water velocity, and substrate with variation of size and deposition, forming different microhabitats.The streams located in the medium and lower basin are characterised by low water velocity, high water temperatures, and substrate composed of fine sediment, detritus and low diversity of microhabitats (Bozzetti and Schulz, 2004).

Assessment of fish community
The fish community was analysed at 34 sampling sites distributed in 24 streams, most of them in the lowland region of the basin.Seven streams had more than one sampling site (Figure 1).Sampling was carried between April 2004 and August 2006 by electric fishing, using a 7 kW generator (model FEG 800 EFKO, Germany) with 750 V direct current at maximum 3 A, due to low water conductivity (between 25 and 90 μS/cm).Fishing effort was 40 minutes, which in most cases corresponded to a stretch of 40 times the mean width of the stream (Angermeier and Smogor, 1995).Common fish species were identified on site and released.Others were fixed in formalin 10% and transferred to ethanol 70% for posterior processing.The scores attributed to each metric were added for each site, resulting in a sum with a minimum of zero and a maximum of 80. Classes were divided in good (scores ≥ 61), moderately impacted (≥41 and ≤60), and impacted (<40), according to Pinto and Araújo (2007).

The Stream Corridor Assessment Survey (SCAS)
Between 2004 and 2006 the Water Committee of the Sinos (Comitesinos) basin organised a SCAS in cooperation with the Universidade do Rio dos Sinos (UNISINOS).SCAS was based on the methodology applied by Yetman (2001) for streams in the Chesapeake basin of the State of Maryland, USA and adapted to local conditions.The survey was carried out by about 200 volunteers from 22 municipalities of the basin.These people were trained to recognise eight impact categories and to attribute severity scores from 1 (small impact) to 3 (severe impact) according to the adapted SCAS manual (Schulz et al., 2004).Volunteers walked along the streams of their municipalities, recorded, photographed and geo-referenced the impacts encountered by GPS.During SCAS a total of 2,000 stream kilometers were assessed and 8,000 field protocols emitted.One of the main products of this study was the cumulative SCAS index, which adds up all impacts encountered in stream segments of five kilometers (Comitesinos, 2009).
The index attributes different weights called impact factors to different impact categories which are multiplied by the severity scores (Table 3).Example: Counted in a five km segment are: 2 sewage pipes (impact factor 5) of severity 2 and one severity 3, one erosion (impact factor number of individuals in a sample, were equally applied in the present study.
Parameters such as total richness, number of species in the water column, of benthic species, of sensitive species, number of dominant species and percentage of carnivorous individuals decrease when degradation increases.However, percentage of Cyprinodontiformes and omnivorous individuals are metrics that increase with degradation, due to the adaptation of the individuals to impacted environmental situations.
Since no undisturbed reference sites were available in the basin, the hypothetic reference of Pinto and Araújo (2007) was used.For each parameter, the reference was the best (highest) or worst (lowest) value obtained among all stations, depending on the ecological significance of the metric (Table 2).
A scoring system from one to ten was applied for all parameters.For example, the highest richness was 28 species, hence considered the reference value.This value corresponds to a score of ten.A richness of 16 species received a score ([16/28] × 10) = 5.7.In the cases where the values of the measurements decreased with a pollution increase, (example: species richness) the best (highest) value was used as reference.For metrics that increase with habitat degradation, the worst value was used as reference, for example: [10-(2.5/87.3x 10)]=9.71where 2.5 is the value of the Cyprinodontiformes percentage of individuals in a stream and 87.3 is the worst of metric value sampled, considered the reference value.25, 30 and 34, where fish were absent.Streams with no fish were located exclusively in urban areas.Linear regression between Shannon-Wiener diversity index and stream order showed a positive significant relationship (R 2 = 0.18; P = 0.01; N = 34), (Figure 2).

Integrity Biotic Index (IBI)
All IBI scores fitted in two impact categories: moderately impacted and impacted.Scores higher than 61, characterising good conditions, did not occur.The highest IBI scores were attributed to moderately impacted streams which occurred predominantly in the middle section of the river basin, which is a transition zone between the headwater and floodplain area.The lowest IBI scores occurred in the lower parts of the basin, where urban and industrial areas are concentrated (Table 5).

Stream Corridor Assessment Survey (SCAS)
Linear regression between SCAS index and stream order did not show a positive significant relationship (R 2 = 0.10; P = 0.56; N = 34).Multiple linear regression between the scores of the seven impact categories and IBI showed a negative significant relationship only for channel modifications (R 2 = 0.53; P = 0.02; N = 34).The same relationship was apparent between channel modification and Shannon-Wiener index, but explaining a lesser degree of variance (R 2 = 0.45; P = 0.02; N = 34).
The impacts present in each section of the streams sampled in this study were obtained by the SCAS realised in the cities between 2004 and 2006.For more details check Schulz et al. (2004).

Statistical treatment
Community diversity was estimated by the Shannon-Wiener diversity index (H') for each sampling site.Simple linear regressions were performed to verify the possible relationship between IBI and SCAS, IBI and stream orders, SCAS and stream order, Shannon-Wiener and SCAS and finally Shannon-Wiener and stream order.The relationships between environmental variables with IBI and with Shannon-Wiener were analysed using a stepwise multiple regression, to evaluate possible effects of these impacts on the biotic integrity of the streams and in diversity.Statistical analyses were performed using the software SYSTAT (version 12.0, 2007).
The Shannon-Wiener diversity index varied between H' = 2.59 in stream number six to H'=0 in stream numbers  Table 4. Abundance and classification of fish species according to the order, family, total abundance, longevity, trophic level, tolerance to disturbances and habit in the water column.c = carnivore, h = herbivore, i = insectivore, o = omnivore; in = intolerant, t = tolerant; b = benthonic, ba = benthonic able to breathe atmospheric O 2 , wc = water column and s = surface.

Discussion
The fish community represented by 61 species was similar to that of a study in the main stem of the Sinos River, where 68 species were recorded (Petry and Schulz, 2006), demonstrating that the diversity of the Sinos River is well represented in the tributaries.
Nine of the 34 sampling sites were considered moderately impacted and 25 impacted.No sampling site achieved an IBI score higher than 61 which was the lower score limit for good conditions.The absence of this highest IBI very different biological characteristics: Species of the genus Astyanax, occupying the water column and feeding on drifting particles, Gymnotus using the riverbanks for shelter and feeding and Hypostomus which occurs on rocky substrate in fast flowing river stretches (Ferreira and Casatti, 2006).Fish assemblages in the streams with the lowest IBI values were composed of Synbranchus marmoratus, Corydoras paleatus and Otocinclus flexillis.These are extremely tolerant species and are the last to disappear in response do environmental degradation (Bozetti and Schulz, 2004).Of these S. marmoratus seems to be the last to die or disappear in poorly oxygenated water bodies.These results indicated that the adaptation of the IBI applying as reference the best (highest) or worst (lowest) value obtained among all stations was effective and can be tested in studies in other regions.The application of the IBI with the SCAS index showed that structural modifications of stream habitat had the most important effect on the IBI.The combination of IBI with quantitative assessments of structural and chemical impacts, as showed in the present study, offers a relatively low cost method for environmental assessment with a rapid return of results.The use of our IBI indicated that this index could be applied in different regions of Brazil, since it is composed by metrics that include functional groups.The only exception as functional groups, is the metric of the presence or absence of Cyprinodontiformes, which is used in tropical regions of the Americas, Asia and Africa (Jaramillo-Villa and Caramaschi, 2005).This metric contains representative indicators of degraded sites, such as species of the family Poeciliidae (Araújo, 1998) and preserved environments, such as native species of Cyprinodontidae (Toham and Teugels, 1999).Cyprinodontiformes is a group widely distributed in Brazil and may be IBI included in studies throughout the country.
These results show that application of the IBI could be an important tool for biomonitoring, when used in addition to other indices, since it infers the quality of environmental prediction using different levels of fish community organization.low and explains only 25% of the variance.Since the SCAS sores are composed of eight impact parameters, it was important to identify the impact category which contributed most to IBI reduction.Multiple regression showed that channel modifications were the most important predictor variable, explaining 53% of the variance in reduction of IBI scores.The number of sewage pipes which are indicators of organic and chemical water pollution and other impact categories that compose the SCAS index did not contribute significantly to the model.According to the results of this study, the decrease of IBI scores in urban areas is principally caused by physical degradation of the aquatic environment.Apparently, canalisation and the inherent homogenisation of microhabitats influence IBI scores more negatively than the decrease in water quality.The Shannon-Wiener diversity index responded as well negatively to canalisation but less apparent.Ferreira and Casatti (2006) observed in a Brazilian stream that structural variables like absence of riparian vegetation, low substrate stability and low water depth, influence biotic integrity more than water quality.Canalised river segments with homogeneous substrate, low water depth and low habitat complexity had a lesser diversity and abundance fish community than undisturbed river stretches (Fausch and Bramblett, 1991;Paller, 2002).
We found correlations between the increase in biotic integrity and diversity with stream order, reflecting the predictive power of this descriptor.Casatti et al. (2006), Braga and Andrade (2005) and Goldstein and Meador (2004) also reported an increase in the number of species per stream order.This relationship of species richness and stream order is considered to be a result of the species area relationship (Angermeier and Schlosser, 1989).Larger streams with higher water discharge sustain a higher diversity of habitats and allow the coexistence of more species with

Figure 2 .
Figure 2. Linear regression between the diversity index of Shannon-Wiener and the stream order for 34 streams in the Sinos River basin.

Figure 5 .
Figure 5. Linear regression between IBI and stream order for 34 streams in the Sinos River basin.

Table 1 .
Pinto and Araújo (2007)d byPinto and Araújo (2007)and the modifications used in the present study.

Table 2 .
Criteria to evaluate the Integrity Biotic Index, using as reference the best (highest) or worst (lowest) value obtained among all stations.

Table 3 .
The different weights called impact factors of index attributes for each impact category.

Table 5 .
Identification of streams moderately impacted and impacted conditions resulting from the application of Biotic Integrity Index.H = Headwaters, M = Middle and L = Lower section.