Diversity and composition of Trichoptera ( Insecta ) larvae assemblages in streams with different environmental conditions at Serra da Bocaina , Southeastern Brazil

Aim: The goal of this study is to examine the composition and richness of caddisfly assemblages in streams at the Serra da Bocaina Mountains, Southeastern Brazil, and to identify the main environmental variables, affecting caddisfly assemblages at the streams with different conditions of land use. Methods: The sampling was conducted in 19 streams during September and October 2007. All sites were characterized physiographically by application of environmental assessment protocol to Atlantic Forest streams and by some physical and chemical parameters. Of the 19 streams sampled, six were classified as reference, six streams as intermediate (moderate anthropic impact) and seven streams as poor (strong anthropic impact). In each site, a multi-habitat sampling was taken with a kick sampler net. The sample was composed by 20 units, each one corresponded to 1 m2 of collected substrate, corresponding 20 m2 of sampling area. The material was placed in a plastic container (500 μm of mesh), washed, homogenized and sub-sampled. For each stream, 6 subsamples were randomly sorted. Results: Were collected 2,113 caddisfly larvae, belonging to 12 families and 28 genera. Hydropsychidae and Leptoceridae were the most abundant families, and Smicridea was the most abundant genus. Sorensen’s index results showed that the streams studied were grouped according to environmental integrity. The Indicator Species Analysis showed only characteristic taxa to reference streams. Canonical Correspondence Analysis showed that caddisfly assemblage was strongly influenced by nitrate concentration, pH and condition of riparian vegetation. Multiple regression analysis indicated significant correlations to five genera with some environmental parameters, besides total abundance of Trichoptera. Conclusions: Ours results showed that degree of environmental impact, mainly the nitrate concentration, pH, and condition of cover vegetation acted as a major factor in determining the Trichoptera assemblages present in the stream of the Serra da Bocaina, separating streams along an environmental gradient.

The aim of the present study was to examine the composition and richness of caddisfly assemblages in streams at the Serra da Bocaina Mountains, Southeastern Brazil, and to identify the main environmental variables, including riparian vegetation, affecting caddisfly assemblages in streams subjected to different conditions of land use.

Study area
This study was carried out in 19 streams of the Serra da Bocaina Mountains.The Serra da Bocaina territory is partly preserved in a conservation unit, the Parque Nacional da Serra da Bocaina (PNSB) with an area of 1 000 km 2 , located between 22°40'-23°20'S and 44°24'-44°54'W), with 60% of native vegetation and the remainder consisting of a 30-y regenerated (secondary) forest.The climate is classified as temperate super humid with annual precipitation of 1,800 mm.Average temperature is around 16 °C but with a high variation between lowland areas (36-38 °C, around 200 m.a.s.l.) to

Introduction
The lotic ecosystems in the Atlantic forest of Southeastern region of Brazil have suffered severe man-induced stress and are threatened by pollution, development of urban areas and loss of riparian vegetation.According to Dean (1997), the Atlantic Forest is known for its high species diversity, the high degree of endemism of its biota.The destruction of this forest has been occurring since the European colonization of Brazil.Deforestation has not only altered physical habitat, but it has also led to increased peak flows and water volumes during flood events in some pasture catchments.These land-use impacts constitute a press disturbance and also can to affect stream community structure and function over the long-term (Harding et al., 1998).
Macroinvertebrate as indicators of human disturbance have a long history for use in evaluating sewage pollution in rivers, among then Trichoptera larvae are often a major component of the invertebrate fauna in lotic ecosystems worldwide (Ward, 1992), corresponding about 8-13% of total abundance.They are fundamental components of the trophic dynamics and energy flow in lakes, rivers, and streams, forming a link between basal resources (organic debris and primary production and secondary consumers such as fishes (Resh & Rosenberg, 1984;Angrisano, 1995;Wiggins, 1996).
Trichoptera, along with Ephemeroptera and Plecotera, is one of the integrant groups of EPT index (Rosenberg & Resh, 1993).The caddisflies stand out in the monitoring of water quality because they have high species richness and abundance, varying levels of sensitivity to physical and chemical changes, and pollution of aquatic ecosystems (Rosenberg & Resh, 1993;Wiggins, 1996).These traits make the order a good indicator of water quality (Collier et al., 1997).

Sampling sites
The sampling was conducted in September and October 2007.In each sampling site, it was conducted a visual environmental assessment protocol (EP) based on the RCE -Riparian Channel Environment protocol (Petersen, 1992).The environmental assessment protocol (EP) was obtained by summing 10 individual metrics based on visual assessment of riparian vegetation cover, stream margins, habitats, substrates or bank conditions.For each parameter a score is given, summed, and the final score is compared to determine one of the five classes of environmental integrity (Appendix 1).
The mean width (in meters) per stream was determined from two sections (10 m apart) perpendicular to the main stream flow.The current velocity was estimated by the head rod method (Waterwatch Australia Steering Committee, 2002), using a stainless steel ruler.In each section, we measured the depth of the stream in meters, (D1, with the thin edge of the ruler into the flow), and (D2, with the flat side face into the flow, creating a standing wave or 'head').These measures were taken at every 30 cm across the stream section.The difference between D1 and D2 is the head.The average head (h) were calculated from these measurements.The average velocity of the stream was determined by the formula: V (m/s) = √ 2 (2x9.81xh),where 9.81 is the gravitational constant.The current velocity of each stream was the average of velocities measured in each sampling year.Stream orders were determined by the study of the tributaries of each stream using the cartographic charts of IBGE (1973aIBGE ( , b, c, 1974 a, b) a, b) (scale 1:50 000).
Before biological sampling, the sites were classified into three levels of impairment: reference sites, moderately impaired sites (intermediate), and severely impaired sites (poor).Reference sites were considered as streams minimally disturbed by meeting located in forested areas, land use with maximum of 20% of the basin area urbanized and ≥ 75% of the upstream basin area forested; width of the riparian zone > 18 m; no visible sign of channelization; and "very good" classification according to the environmental assessment protocol.Major impacts at 'impaired' sites were removal of riparian vegetation and alterations of physical characteristics of streams with 'intermediate' or 'poor' classification.For the 'intermediate' condition, the following a priori conditions should be met: deforestation of 50-70% of the upstream area; silting in riffle mesohabitats covering 30-50%.
In each site, a multi-habitat sampling was taken with a kick sampler (500 µm mesh), the sampling was proportional to substrate availability in the stream stretches studied.The sample was composed of 20 units, corresponding to total sampling area of 20 m 2 substrates, and each one corresponds to 1 m 2 of collected substrate.The sample was combined and preserved in 80% ethanol.In laboratory, the sample of each stream was washed, homogenized and placed in a plastic container with total size 64 x 36 cm, divided into 24 quadrats of aluminum with mesh of 500 µm.Each quadrat measures 10.5x8.5cm, with an area of approximately 90 cm 2 .For each stream, 6 subsamples were randomly removed (representing each subsample one quadrat) that were sorted in white trays (Oliveira et al., 2011).Caddisfly larvae were identified with aid of the taxonomic keys by Angrisano (1995), Pes et al. (2005) and Wiggins (1996).

Data analysis
Community diversity and evenness were calculated using the Shannon-Wiener and Pielou indexes, respectively (Elliot, 1977;Ludwig & Reynolds, 1988).Taxonomic richness was estimated as the total number of different taxa found in each sample and by rarefaction method (Gotelli & Colwell, 2001) using the program Past, version 1.40 (Hammer et al., 2001).
Sorensen's Similarity Index was used for analyzing similarities between taxonomic composition of Trichoptera fauna based on a presence-absence matrix for the genera.The Sorensen's Index matrix was submitted to a cluster analysis through the average association method (UPGMA).Characteristic groups of each stream type were determined through Indicator Value Method -IndVal (Dufrêne & Legendre 1997).This method matches information on species abundance and frequency of occurrence among groups.A Monte Carlo permutation test was employed to test significant associations of taxa and groups of sites (p < 0.05).
A Canonical Correspondence Analysis (CCA) was used to determine the factors that might be influencing the caddisfly fauna abundance and distribution.The total abundance data of all streams were log-transformed (n + 1).Environmental features included were: pH, Conductivity, width, depth, velocity, discharge, Dissolved Oxygen, Ammonia, Nitrate, Nitrite, Total Alkalinity, fecal coliform, besides the degree preservation of riparian vegetation (RV) using the score of EA item 9. CCA, Sorensen's Similarity index and Indicator Value Method were performed using PC-ORD program version 4.14 (Mccune & Mefford, 1999).
Multiple regression analysis was performed to evaluate the relation between environmental features (independent variable) to species distribution and abundance of the same (dependent variable).The dependent variables considered in this analysis were abundance of each taxa in each site, total abundance and total richness.This analysis was performed using STATISTICA 7.0 (Statsoft, 2004).

Environmental parameters of streams
Of the 19 streams sampled, six were classified as reference (score > 16), and the other streams were classified as impaired, being six streams as intermediate, with moderate anthropic impact (score > 9.0 < 16) and seven streams as poor, strong anthropic impact (score > 0 < 9.0) (Table 1).The environmental parameters recorded from the studied streams are shown in Table 2. Temperature values do not varied significantly among the streams.Stream S03 show the lowest pH values (5.70) while S17 the highest (8.04).The stream S06 classified as poor, showed the highest values of Conductivity (140.8µS.cm - ), Ammonia (> 2.5), and Alkalinity (123.2 mg/L).The values of dissolved oxygen (DO) do not showed significant variation because some poor streams obtained high values.

Trichoptera community composition
A total of 2,113 individuals of Trichoptera larvae belonging to 12 families and 28 genera were collected.Hydropsychidae and Leptoceridae were the most abundant families with 916 and 535 specimens respectively.The taxonomic richness (rarefactions and observed), Shannon's diversity, Pielou's evenness index, and total abundance are shown in Table 3.The richness in reference streams was standardized for 63 individuals, intermediate for 9 individuals and poor streams for 18 individuals by the rarefaction richness.This value represents the stream with lowest abundance.Although, the S06 and S08 had lower abundance values than site S02, these two sites were not considered in rarefaction analysis because both had only one genus with one and four individuals respectively.The highest values of richness and diversity were found in sites S18 and S19 both located in preserved areas.
Cluster Analysis results based on the Sorensen's index (Figure 2) show the presence of three principal  Intermediate and poor streams do not showed characteristic taxa.
In the CCA the three first axis explained 0.350, 0.280, and 0.230 respectively of variation, and were strongly influenced by nitrate concentration, pH and condition of riparian vegetation , separating streams with good ecological condition from impaired streams (Figures 3 and 4).

Discussion
Biodiversity loss in freshwaters is a mounting threat from widespread human disturbances, and benthic macroinvertebrates are key indicators in determining patterns of stream ecosystem degradation (Pond, 2011).In our study, we observed a great loss of diversity in intermediate and poor streams when compared with the reference ones.The low diversity of Trichoptera at impaired sites is related to loss of habitats by urbanization and deforestation, causing the increase of sedimentation on rocky substrates and reduced litter input in the stream (Allan et al., 1997;Townsend et al., 1997;Bispo & Oliveira, 2007).Hydropsychidae and Leptoceridae were the main families found in this study.These data confirm the greater abundance of these families in Tropical regions, as reported by Flint et al. (1999).According Bonada et al. (2004), Hydropsychidae is regarded as a very tolerant family all over the world, with species being segregated within different water quality characteristics along the river.In our study, Smicridea was the most abundant in all sites, mainly in the impaired sites, where represented more than 60% of the specimens.This result denotes a high tolerance of this genus to small impacts as absence of riparian vegetation, sedimentation, or small input of organic matter, and corroborates the results found by Righi-Cavallaro et al. ( 2010) and Massoli & Callil (2014) for other regions of the country.According Bentes et al. (2008), Smicridea is a water quality bioindicator with generalist habit which can be found in environments with different degrees of preservation.Smicridea larvae are found in rocky substrates in moderate or high current velocity and its main source of food is fine particulate organic matter (FPOM) that is trapped in the capture nets they build (Wiggins, 1996;Merritt & Cummins, 1996).
The results of Sorensen's index showed the presence of different communities, characteristic to each degree of environmental impact.According to Allan (2004) the land use patterns are considered one of main determining factors in macrofauna distribution.According Bonada et al. (2004), certain caddisfly families and species are sensitive to some variables but more tolerant to others, making the Trichoptera good tools for biomonitoring.Atopsyche, Helicopsyche, Phylloicus and Triplectides were characteristic to reference streams, corroborated by results of multiple regression for caddisfly total abundance that showed positive correlations with preservation of riparian vegetation.Helicopsyche, Phylloicus and Triplectides are primarily leaf eaters (shredders or scrapers) having strong association with degree of preservation of riparian zone, that provides the process of determining taxon and assemblage tolerance.
Our study represents first investigations about assemblages of Trichoptera larvae and environmental quality in Serra da Bocaina streams, and confirms previous studies conducted to same area about environmental quality, that degree of environmental impact, mainly the condition of cover vegetation and amount of fecal coliform, act as a major factor in determining the distribution of Trichoptera assemblages present in the stream.The identification of which environmental factors were influencing the composition of aquatic insect fauna is very important to guide programs of assessment and conservation of biodiversity in rivers of Southeastern Brazil.
food and material for shelters.Nogueira et al. (2011) found that species highly dependent to leaf bags as Phylloicus, showed significant declines in impaired streams or without riparian vegetation.Pereira et al. (2012) found Helicopsyche and Phylloicus as good indicators of preserved environments.Atopsyche larvae are predators and their main food items are larvae of Diptera, Trichoptera, Ephemeroptera and Oligochaeta (Reynaga & Rueda Martín, 2010).Ours results could prove that some taxa may be sensitive to changes in the quality of the stream and riparian vegetation, and might be good candidates for using as indicator taxa in bioassessment programs.
Canonical Correspondence Analysis showed that degree of environmental impact, directed by the change in nitrate concentration, pH and condition of riparian vegetation, were the main influencing factors that separated the stream along with an environmental gradient.Strieder et al. (2006) observed that the concentration of nitrate and nitrite had leading role in the distribution of species of Simulium in rivers in the Rio Grande do Sul state separating streams according to the degree of environmental impact.Hepp et al. (2013) studying EPT distribution in urban streams with different environmental impacts found that BOD, nitrate and phosphorous affected particularly Trichoptera and Plecoptera.Nogueira et al. (2011) observed that the flow, vegetation and conservation of the environment were determining factors in the species composition at the Suiá-Miçú River basin, Mato Grosso, Brazil.Allan et al. (1997) found that land use was a strong predictor of biological and habitat integrity.The riparian vegetation plays an important role in the dynamic of water bodies, influencing directly or indirectly in the process and system of community (Vannote et al., 1980).Also, a higher amount of sediment was found on the stones as a consequence of the large areas of land recently denuded.Rodrigues-Filho et al. (2015) analyzing the alterations in land uses based on amendments to the Brazilian Forest Law and their influences on water quality of a watershed found that even the suppression of only 20% of vegetation cover in the watershed is not compatible with sustainable practices and would result in losses of important ecosystem services and increase of loading nutrients in watersheds.According Whittier & Van Sickle (2010) quantifying an overall human disturbance gradient is probably the least tractable part of Appendix 1. Environmental Assessment Protocol to Atlantic Forest adapt from RCE (Petersen, 1992).

Figure 1 .
Figure 1.Map of Southeastern Brazil region highlighting the Southern of Rio de Janeiro state and the North of São Paulo state with the sampling sites.

Figure 2 .
Figure 2. Cluster Analysis (UPGMA method) based on Sorensen's Index values for 19 streams in Serra da Bocaina Mountain.R -Reference sites; I -Intermediate sites; P -Poor sites.

Table 2 .
Environmental features measured for 19 rivers studied in Serra da Bocaina Mountains.

Table 1 .
Localization, score of Environmental Assessment Protocol (EP) and Environmental classification for each studied stream in Serra da Bocaina Mountains.

Table 3 .
Trichoptera taxa collected in 19 streams in Serra da Bocaina Mountains, Southeastern Brazil.

Table 4 .
Taxa with significant values of Indication (p < 0.05, 1000 permutations) for each site group (reference, intermediate and poor streams) studied at Serra da Bocaina Mountain, Southeastern Brazil.Where Max Group 1: reference; max group 2 -intermediate and max group 3 -poor.