Preditores ambientais, estrutura espacial e distribuição geográfica potencial de macrófitas aquáticas invasoras no Rio São Francisco, Brasil

Abstract

The geographic distribution modelling is an important tool, which has been used for the management of invasive species. The modelling creates maps that indicate environmentally suitable areas for the occurrence of target species. Such maps are the result of optimization procedures (correlations between the occurrence records of the target species and environmental variables). Studies that analyze explanatory environmental variables about the patterns of occurrence of invasive species have also contributed to the control of invasive plants. However, for certain biological communities, for instance macrophytes, studies with such approaches still lack methodological investigations with critical evaluation. This thesis comprises two chapters that discuss major methodological paradigms about the studies of environmental predictors and potential geographic distribution of aquatic plants, using invasive species as study models. The first chapter refers to a critical analysis of the potential geographic distribution modelling of aquatic macrophytes, analyzing two invasive species, Egeria densa and Eichhornia crassipes, in the São Francisco River, northeastern Brazil. This research evaluates: differences between the modelling of floating and submerged fixed macrophytes; best records of the species (field records or herbarium/article information) or environmental predictors (Bioclimatic or limnological) in such models; the influence of the depth of collection of limnological data on the performance of the models; and the importance of statistic treatment of environmental predictors during optimization process. The models were computed after optimization process performed using the Maxent algorithm, and binomial tests were used to compare the efficiency of the models. The models for E. crassipes (floating species) showed veracity errors on the recorded points of occurrence (regardless of the data source), which is probably due to its capacity to drift in environments with water flow, causing such methodological bias. Optimized models with field records, limnologic predictors collected on the surface and/or with selected non-collinear predictors performed better than those generated using records obtained from herbaria/articles, with bioclimatic or limnologic predictors collected on the deep, and/or without statistical selection of environmental predictors. The second chapter analyzes the presence of spatial structure in the occurrence of E. densa in semi-lotic regions of the São Francisco River. Also, we identified and partitioned the explanation of the set of environmental predictors (limnology or bioclimatic) or of the space factor on its occurrence pattern. Between 2006 and 2010, we conducted 12 expeditions to collect limnologic and occurrence data of the analyzed species at 68 sampling points. Additionally, the information of 19 bioclimatic variables were collected from BioClim. We performed multivariate analysis (NMDS) and t test to evaluate the adjustment of E. densa to semi-lotic environments of the river, and used correlograms to test spatial autocorrelation in the response variable. We applied automatic selection of models to identify the set of environmental predictors and spatial filters that explain both the patterns of occurrence of E. densa and the partition of variance. The results confirmed the adjustment of E. densa for lentic environments, and the presence of space structure. The most parsimonious model indicated that conductivity, inorganic phosphate, total phosphorus, transparency, turbidity, and pH are the main environmental predictors of the occurrence of E. densa for semi-lotic environments in river. Selected models show that the explanation attributed to the space factor was greater than that of environmental predictors, and that physical barriers in these ecosystems limited the dispersion of E. densa. In short, this thesis provides an unprecedented critical analysis of the methodological and conceptual issues concerned to the potential geographic distribution or environmental predictors of invasive aquatic macrophytes, allowing new insights into the knowledge and management of such plants.