Responses of freshwater molluscs to environmental factors in Southern Brazil wetlands

Freshwater molluscs play an important role in aquatic ecosystems, providing food for many fish species and vertebrates. Investigations on factors that determine mollusc species richness and distribution in wetland systems are scarce in the Neotropical region. The main goal of this study was to determine how much variation in mollusc richness and composition is explained by area, hydroperiod, altitude, water conductivity, and dominant aquatic vegetation. This survey was performed in an extensive area of a Neotropical region (~280,000 km in southern Brazil), with a large number of wetland systems (111) and covering a wide gradient of altitude and wetland surface area. The mollusc richness was positively associated with wetland area and negatively associated with altitude. The richness and composition of the freshwater molluscs were similar between permanent and intermittent wetlands and it did not differ significantly between aquatic bed and emergent wetlands. The first three axes of CCA explained 16.2% of the total variation in the composition of the freshwater mollusc observed. The variation in the composition had a correlation with wetland area, altitude and water conductivity. Our results showed that the wetlands are important habitats for molluscs in southern Brazil, and that the richness and the composition of molluscs were associated with area, altitude, water conductivity and dominant vegetation.

The environmental factors that determine the mollusc richness and distribution in inland waters have been investigated.Water hardness, conductivity, pH altitude, substrates (plants), food resources (plants) and biotic interactions have been reported as important environmental predictors of molluscs (Dillon, 2000;Brown, 2001;Gérard, 2001;McMahon and Bogan, 2001;Heino and Muotka, 2006;Sturm, 2007;Horsák et al., 2007).The species-area relationship was not always consistent for wetland molluscs.While the richness of snails tended to increase with wetland area (Lassen, 1975;Aho, 1978;Brönmark, 1985), this relationship was not significant for Sphaeriidae (Oertli et al., 2002).Investigations on factors that determine mollusc species richness and distribution in wetland systems are scarce in the Neotropical region.
The main goal of this study was to determine how much variation in mollusc richness and composition is explained by area, hydroperiod, altitude, water conductivity, and dominant aquatic vegetation.This survey was performed in an extensive area of a Neotropical region (~280,000 km 2 in southern Brazil), with a large number of wetland systems (111) and covering a wide gradient of altitude and wetland surface area.

Introduction
Wetlands are important sites for biological conservation because they support a rich biodiversity and present high productivity (Mitsch and Gosselink, 2000).However, biodiversity in wetlands is being reduced in most of the world by agricultural, urban and industrial development (Shine and Klemm, 1999).Almost half of the world's wetlands have disappeared over the last century due to agricultural and urban development (Shine and Klemm, 1999).The rapid degradation of wetlands and the insufficient status of scientific knowledge on patterns of the species richness bring the urgent need for ecological studies to provide scientific support to biodiversity management and conservation programs.
One of the main hydrological characteristics of South America is the existence of large wetlands (Neiff, 2001).Approximately 50% of the inventoried wetlands in South America are located in Brazil (Naranjo, 1995).Conservative data indicate that approximately 90% of the wetlands in southern Brazil disappeared in the last century due to agricultural expansion, especially rice and soy plantations (Gomes and Magalhães, 2004).In this sense, the understanding of species composition and richness patterns in fragmented and natural wetlands is a priority for biodiversity conservation strategies and land/ water management in southern Brazil.
The macroinvertebrate community is an important trophic level in wetland systems, providing food for several wildlife species.The macroinvertebrate structure in wetlands has been associated with area and habitat heterogeneity (Oertli et al., 2002;Batzer et al., 2004;Panatta et al., 2006;Studinski and Grubbs, 2007;Stenert et al., 2008); hydroperiod (Tarr et al., 2005;Whiles and Goldowitz, 2005;Stenert and Maltchik, 2007); altitude (Oertli et al., 2002;Jacobsen, 2004); water chemistry (Heino, 2000;Batzer et al., 2004), and water depth and temperature (Zimmer et al., 2000;Hall et al., 2004;Tarr et al., 2005;Studinski and Grubbs, 2007).However, most of these surveys were developed in the North Hemisphere and Australia.In the Neotropical region, Stenert and Maltchik (2007) demonstrated that area and hydroperiod are important predictors of macroinvertebrate richness and composition in 72 wetlands distributed in southern Brazil and the occurrence of molluscs was observed in more than 65% of studied wetlands.
Freshwater molluscs play an important role in aquatic ecosystems, providing food for many fish species (McMahon and Bogan, 2001;Garcia et al., 2006;Fagundes et al., 2008) and vertebrates (Cummins and Bogan, 2006).A total of 373 native species of freshwater molluscs were registered in Brazil, 117 attributed to bivalves and 256 to gastropods, but these numbers are underestimated and may represent only 50% of the total malacofauna (Simone, 2006).Additionally, about nine alien species have been recorded in Brazilian inland waters (Simone, 2006), but accurate estimates of the number of mollusc species in southern Brazil are unknown.Mollusc collections were carried out using a kick net (D-shaped, 30 cm width, 400 µm mesh).Sampling was limited to the littoral zone of wetlands (water depths of less than 50 cm), by kicking up the substrate and then sweeping above the disturbed area.The sampling effort was the same for all wetlands, represented by 25 sweeps of 1 m over the various habitats of the littoral zone (detritus, rooted macrophytes and other dominant vegetation).Sweeps were pooled into one sample per wetland (3.5 L plastic bucket) and preserved in situ with 10% formaldehyde.
In the laboratory, each sample was washed through a 400 µm sieve, and leaves, stems, and other debris were removed.The resulting material was preserved with 80% ethanol.Individuals were identified at genus and species level (when possible), by using the extensive malacological literature available.Ancylids and sphaeriids were identified only at family level due to the shell preservation condition.Mollusc richness corresponded to the number of taxa (genera or species) collected in each sampled wetland.

Data analyses
The continuous environmental variables considered in the analyses were wetland area, altitude, and water conductivity, and the categorical variables considered in the analyses were wetland hydroperiod (permanent and intermittent) and vegetation cover (aquatic bed and emergent).In order to remove the heteroscedasticity, continuous environmental variables were log transformed.
Correlations among area, altitude and water conductivity were tested using the Pearson's correlation coefficient.Variation of mollusc richness between permanent and intermittent wetlands was quantified through a t-test.In permanent wetlands, richness variation between aquatic bed and emergent vegetation was also analysed through a t-test.Multiple regressions (GLM) were carried out to ascertain the extent to which wetland area, altitude and water conductivity explained the patterns of mollusc richness in all wetlands and in permanent and intermittent wetlands.
The direct multivariate analysis of the environmental parameters in the composition of freshwater molluscs in the studied wetlands was carried out through Canonical Correspondence Analysis (CCA) (Ter Braak, 1986), using PC-ORD Version 4.2 (McCune and Mefford, 1999), and the significance of the axes generated in the analysis was validated through the Monte Carlo test (using 5,000 iterations) (Ter Braak and Šmilauer, 1998).Environmental variables utilised in the CCA were: wetland area, altitude, water conductivity, and the combination of hydroperiod (permanent and intermittent) and dominant vegetation (aquatic bed, emergent and multi-stratified).Environmental data were centred and normalised.Biological variables were based on the presence and the absence of the mollusc taxa occurring in four or more wetlands.Composition variations between permanent and intermittent wetlands, and among aquatic ture is lower than 10 °C in the winter, and the maximum temperature is higher than 32 °C in the summer (RADAMBRASIL, 1986).
Rio Grande do Sul has approximately 3,441 wetlands, with a total inundation area of approximately 30,332 km 2 (Maltchik, 2003), distributed in five geomorphologic provinces: Coastal Plain, Central Depression, Crystalline Shield, Highlands and Pampas (Hausman, 1995).The vegetation is characterised by small fragments of forest, with temperate and tropical grassland areas.The forested areas are represented by temperate summer-green and mixed ever-green deciduous forests, and temperate mountainous coniferous forest.The grasslands are represented by savanna, steppe, and pioneering formations (RADAMBRASIL, 1986;Rambo, 2000).
A total of 111 wetlands were sampled in the state of Rio Grande do Sul based on three criteria: 1) area no larger than 10 ha; 2) macrophyte presence, and 3) fairly even distribution of the wetlands across the state of Rio Grande do Sul (Figure 1).Sampling was performed initially in the eastern portion of the state of Rio Grande do Sul, and then moved towards the western portion.Each wetland was sampled once from March to October 2002, always during the period with surface water.The studied wetlands were classified in permanent (84 wetlands) and intermittent (27 wetlands).Permanent wetlands retain water for the entire hydrological cycle, whereas intermittent wetlands eventually dry up, retaining water for at least four months of the year (Maltchik et al., 2004).All wetlands presented vegetation cover higher than 30% of the total wetland surface.While permanent wetlands were represented mainly by aquatic bed (submerged and floating plant species -45 wetlands) and emergent (erect herbaceous vegetation -29 wetlands) vegetations, intermittent wetlands showed predominance of emergent vegetation (24 wetlands).
The wetland area was measured in the field.The wetland boundaries were determined based on 1) visual observations of the watermarks, drift line and/or owners' information, and 2) vegetation indicators (e.g., plants with morphological, physiological or reproductive adaptations to prolonged saturation/inundation, and the proportion between aquatic and terrestrial species in the plant community).The wetlands location and altitude were determined using a GPS receiver (model GPS III Plus, Garmin).

Data collection
Two surface water samples were collected in the littoral zone of each wetland using polyethylene bottles (500 mL).All samples were placed on ice in dark containers and taken to the laboratory, and immediately filtered upon return to the laboratory (Whatman ® GF/F glass fiber filters, pore size 0.7 µm).Analysis of water samples did not exceed 3-4 days after field collection.Measurement of water conductivity (µS.cm -1 ) was performed with the methodology provided by APHA (1989).
bed, emergent and multi-stratified wetlands were tested through the Multi-Response Permutation Procedures (MRPP).The MRPP is a multivariate analysis which tests differences between two or more groups.

Results
The range and mean values of wetland area, altitude, and water conductivity in the studied wetlands are shown in Table 1.Wetland altitude was negatively correlated with wetland area (r = -0.486,P < 0.001) and water conductivity (r = -0.274,P = 0.004).Wetland area was not correlated with water conductivity (r = 0.126, P = 0.187).
A total of eight freshwater mollusc families and 10 genera were found in the studied wetlands (Table 2).
Table 2. Taxonomic list of freshwater molluscs and the number of wetlands where each mollusc taxon was found according to hydroperiod and dominant aquatic vegetation (PA = permanent wetlands with aquatic bed vegetation; PE = permanent wetlands with emergent vegetation; PM = permanent wetlands with multi-stratified vegetation; IE = intermittent wetlands with emergent vegetation; and IM = intermittent wetlands with multi-stratified vegetation).

Taxons
Hydroperiod and vegetation

Discussion
The observed mollusc richness represented the species molluscs usually found in lentic environments of southern Brazil, Argentina and Uruguay.The most frequent taxa (Biomphalaria, Pomacea canaliculata and Drepanotrema) was previously observed in different wetland classes, such as lakes, ponds, streams, backwater rivers, dams and rice fields (Chieffi and Moretti, 1979;Bonetto et al., 1982;Mochida, 1991;Pereira et al., 2000aPereira et al., ,b, 2001;;Kloos et al., 2004;Agudo-Padrón and Oliveira, 2008).The high frequency of Pomacea canaliculata, Pisidium and Eupera can be attributed to their ability to support variations at water level (Bachmann, 1960;Junk andRobertson, 1997, Kretzschmar andHeckman, 1995).Gundlachia, Omalonyx convexus and Eupera were also observed floating in water in Neotropical wetland systems (Neiff and Neiff, 2006).The scarcity of Heleobia should be investigated, since this genus was commonly associated to aquatic macrophytes in this region (Coimbra et al., 2005).Darrigran (2002) pointed out that the introduction of alien species was responsible for the declining of Heleobia species in Argentina.The absence of the Mycetopodidae (Anodontites e Mycetopoda) and Hyriidae (Diplodon genus) families may be associated to the kind of sampler used, since these individuals are usually observed buried in deep sediment.Mansur and Pereira (2006) observed eight bivalve species in southern Brazil wetlands.
A positive relationship between macroinvertebrate richness and area has been found for specific taxonomic groups of macroinvertebrates (Spencer et al., 1999;Heino, 2000;Oertli et al., 2002;Sanderson et al., 2005;Studinski and Grubbs, 2007).Stenert and Maltchik (2007) showed that area was an important predictor of macroinvertebrate richness and composition in southern Brazil wetlands.Our study showed that the wetland area was positively associated with mollusc richness.However, this relationship is not always consistent, Mollusc richness ranged from 0 to 8 per wetland, and it was positively associated with wetland area and negatively associated with altitude (GLM, R 2 = 0.103, F 3,107 = 4.113, P = 0.008).The richness of the freshwater molluscs was similar between permanent and intermittent wetlands (t 109 = 0.770, P = 0.443), and water conductivity was not associated with mollusc richness (P > 0.05).In relation to dominant aquatic vegetation, the richness did not differ significantly between aquatic bed and emergent wetlands (t 72 = -0.587,P = 0.559).
While the mollusc composition was similar between permanent and intermittent wetlands (MRPP, A = 0.0005, P = 0.365), the molluscs composition was different between aquatic bed and the multi-stratified wetlands (MRPP, A = 0.082, P = 0.002) and between emergent and multi-stratified wetlands (MRPP, A = 0.049, P = 0.003).The first three axes of CCA explained 16.2% (7.7% in axis 1, 5.4% in axis 2, and 3.1% in axis 3) of the total variation in the composition of the freshwater mollusc (Table 3).Wetland area, altitude and permanent wetlands with multi-stratified vegetation were correlated with axis 1. Water conductivity and permanent wetlands with emergent vegetation were correlated with axis 2. The Monte-Carlo simulation test (5,000 iterations) showed that the variation in the mollusc composition had a correlation with both variable sets correlated with axes 1 and 2 (Table 3).While Pomacea canaliculata, Pisidium, Eupera and Ancylidae were more frequent in permanent wetlands with larger area and located at lower altitudes, Omalonyx convexus, Lymnaea columella, Drepanotrema and Biomphalaria were observed mainly in wetlands with lower area and located at higher altitudes.Omalonyx convexus, Pisidium and Eupera were more associated with more elevated water conductivity (Figure 2).Omalonyx convexus was also associated with permanent wetlands and emergent vegetation (Figure 2).Ancylidae and Biomphalaria were observed mainly in wetlands with lower water conductivity (Figure 2).was negatively associated with wetland altitude in this region.Our study showed that the mollusc richness was negatively associated with altitude.The lower richness of molluscs observed in high altitudes may be related with a low capacity of some species to tolerate extreme environmental and climatic conditions (see Hausdorf, 2006).Biomphalaria, Drepanotrema and Lymnaea columella were more frequent in the studied wetlands located at higher altitudes.The pulmonate genera have a strong ability to support harsh conditions (Brown, 2001).Sturm (2007) also observed a higher occurrence of Planorbidae species and other generalist molluscs in lakes of high altitude in the Eastern Alps.The biology of the amphibious gastropod Omalonyx convexus is poorly known, making since some small studied lakes also had a great number of mollusc species.Oertli et al. (2002) did not observe the relationships between wetland area and Sphaeriidae richness.
The macroinvertebrate richness usually declines from low to high altitudes, although this pattern can be blurred from lowlands to midlands (ca.500 m) (Stoneburner, 1977;Miserendino, 2001;Jacobsen, 2004).Oertli et al. (2002) regarded altitude as an important richness predictor to some macroinvertebrate families in Swiss ponds.While Stenert and Maltchik (2007) did not observe a negative relationship between altitude and macroinvertebrate richness in southern Brazil wetlands, Panatta et al. (2007) showed that the richness of Chironomidae ( Barker, 2001) (herbivore) inhabits emergent vegetation and has been commonly found near the margins of many aquatic environments (Thomé et al., 2006).
Our results showed that the wetlands in southern Brazil are important habitats for molluscs, and that the richness and the composition of molluscs are associated with area, altitude, water conductivity and dominant vegetation.In terms of conservation, a wetland does not necessarily have to be large and located in lower altitudes to have a high value.In this study, smaller wetlands at high altitudes supported a mollusc composition that differed from those of larger wetlands and located in lower altitudes.In this sense, there is a need to promote conservation of all wetland systems in southern Brazil, regardless of their area, altitude and kind of dominant vegetation.These points should be seen as important to determine the environmental factors that shape and maintain biodiversity in these ecosystems.Such information is essential to develop conservation and management programs for wetlands in this region, where more than 90% of the wetland systems have already been lost, and the remaining ones are still at high risk due to the expansion of rice production and exotic Eucalyptus and pine plantations.
it difficult to explain its association with wetlands of high altitude.
Several studies suggested that the hydroperiod influences the structure of wetland macroinvertebrate communities (Wellborn et al., 1996;Zimmer et al., 2000).Our study showed that the mollusc richness and composition were similar both in permanent and intermittent wetlands.The majority of the genera observed (pulmonate gastropods) are able to deal with long term hypoxia (Brown, 2001).The aestivation process was observed in several species of Biomphalaria, Drepanotrema and Lymnaea columella, including B. tenagophila (Teles and Marques, 1989) in Brazil.Many bivalve populations are restricted to shallow waters and susceptible to droughts or dry periods.However, the bivalves observed were represented exclusively by Sphaeriidae, a family with high tolerance to air exposure (McMahon and Bogan, 2001).
Water conductivity is closely associated with diversity and richness of aquatic invertebrates (Friday, 1987;Nyman et al., 2005).Lewis and Magnuson (2000) and Metzeling et al. (2006) found strong relationships between mollusc richness and electrical conductivity.Our results did not show any association between water conductivity and mollusc richness.Many studies have shown that this relationship is not a consensus regarding wetland systems.Horsák (2006) remarked that the snail richness did not increase with electrical conductivity.Several studies with Pisidium and Lymnaea columella have found different results in relation to the association between mollusc richness and water conductivity (Marchese and Drago, 1992;Berezina, 2003;Kazibwe et al., 2006;Horsák, 2006;Sturm, 2007).Our results showed that Biomphalaria and Ancylidae were mainly observed in wetlands with lower water conductivity.Berezina (2003) observed that Lymnaea columella is usually more tolerant to low values of salinity, and Kazibwe et al. (2006) sustained that Biomphalaria tend to avoid waters with high conductivity.Pisidium was registered preferentially in areas with high electrical conductivity in South America (Marchese and Drago, 1992).

Figure 1 .
Figure 1.Location of study area (state of Rio Grande do Sul, Brazil) with the position of all studied sites.

Figure 2 .
Figure 2. Diagrams of Canonical Correspondence Analysis (CCA) ordination: a) Relationship among mollusc taxa and environmental variables (vectors); and b) Relationship among environmental variables (vectors) and wetlands categorised according to dominant vegetation.Abbreviations: PM wetlands = permanent and multi-stratified wetlands, and PE wetlands = permanent and emergent wetlands.

Table 1 .
Ranges and mean of selected environmental variables characterising the studied wetlands in southern Brazil.

Table 3 .
Summary of Canonical Correspondence Analysis (CCA) results, for the 111 wetlands studied in southern Brazil.