Modelagem da adequabilidade de hábitat de Corbicula fluminea (Bivalvia, Corbiculidae), Pimelodus pintado e Parapimelodus nigribarbis (Siluriformes, Pimelodidae) e Loricariichthys anus (Siluriformes, Loricariidae) em relação a preditores ambientais no Lago Guaíba, RS, Brasil

• Autor(a) principal: Silveira, Thiago Cesar Lima
• Data de Publicação: 2015
• Tipo de documento: Tese
• Idioma: por
• Título da fonte: Biblioteca Digital de Teses e Dissertações da PUC_RS
• Texto Completo: http://tede2.pucrs.br/tede2/handle/tede/6156
• topic: ZOOLOGIA; ICTIOLOGIA; PEIXES - RIO GRANDE DO SUL; PEIXES - ESPÉCIES; SILURIFORMES; HABITAT DOS ANIMAIS; CIENCIAS BIOLOGICAS::ZOOLOGIA

The objective of this study was to model the habitat suitability for the bivalve Corbicula fluminea, and the fishes Pimelodus pintado, Parapimelodus nigribarbis and Loricariichthys anus in relation to environmental variables in Guaíba Lake, Rio Grande do Sul, Brazil. Guaíba Lake, a large water body beside Porto Alegre city with approximately 50 km long and 19 km wide at the widest point. It has north to south pollution gradient, different sediment characteristics and habitat configurations in its margins. Samples (n = 54) of C. fluminea were collected with Ekman dredge (10 subsamples, sampled area=2225 cm2) and the fish sampled with gill nets with different mesh sizes exposed during 20 hours at the same sampling points. We used Generalized Additive Models (GAM) to fit a model for C. fluminea abundances. The variables evaluated for the model fitting comprise the distance from streams and channels, the distance in meters from north to south, depth, water velocity, amount of thermotolerant coliform (TC) and sediment characteristics. The model selected for C. fluminea (r2 = 0.55) used depth (p = 2-16) and the diversity sand grain sizes (p = 2.52-15) as predictors. For fish modeling, three approaches were evaluated (GAM), Boosted Regression Trees (BRT) and Generalized Linear Modeling (GLM). GLM was the method more accurately among the three methodologies for the three species: P. pintado, r2 = 0.60, Root Mean Squared Error (RMSE) = 4.71; P. nigribarbis, r2 = 0.84, RMSE = 6.77; and L. anus, r2 = 0.72, RMSE = 2.40. The most important predictors in the model for P. pintado were related to the sediment and current speed. The model selected for P. nigribarbis had the most important variables: depth, distance from margins and CT. The adjusted model for L. anus had the most important predictors CT and distance from margins. The results suggest that the distribution of C. fluminea, P. pintado, P. nigribarbis and L. anus is not random, though structural and environmental predictors influence the habitat occupation of these organisms. Furthermore, the results show the spatial distribution of species assessed on a small-scale, concerning the majority of spatial distribution studies. Thus, the findings contribute to a better understanding of the autoecology for those species, allowing the prediction to future scenarios in Lake Guaíba, constituting as a tool for environmental management and conservation actions.