Pteridófitas de Santa Catarina, Brasil: diversidade, distribuição geográfica e variáveis ambientais

Abstract

Santa Catarina is one of the few Brazilian states with a well known flora. It is thereforepossible to carry out further data analyses without additional field work. In this study, we analyzed the variation in species composition of ferns and theenvironmental variables that may be influential, so as to indicate the endangered species. Thepreparation of the database was carried out using herbarium specimens, data from scientific collectionsindexed by the SpeciesLink network and on data from the Atlantic Forest Project. The state territory was divided into 50 x 50 km cells, called sample units. 60 sampling units were selected for the study area. For all sample units, species richness by Jackknife method was computed. In further statistic analyses only sample units with less than 60% of estimated richness were used. The WPGMA e TWINSPAN analyses were performed, as well as ordinations by CCA and DCA, with the use of climatic variables selected by VIF. Then best fitting models were selected by multiple regression based on minor AICc for predicted species composition on DCA axes. The selection of the definitive model considered AICc less than 2. D'Agostino-Persons test appointed normality of residuals of both axes. 439 infrageneric taxons of pteridophytes were registered, with 20 new citations for the state. InDense Ombrophylous Forests occur 393 spp., Mixed Ombrophylous Forests 274 spp. and in Seasonal Deciduous Forests 73 spp. Seven species were considered vulnerable and one (Asplenium lacinulatum) was considered critically endangered. Seven species were classifiedas extinct and data availability was deficient for some species. Of the total of 439 species, 65 (14.8%) were sampled in only three sites. 39.7% of all species occur in more than 10 sample units, i.e. in an area larger than 25,000km² (25% of the states territory). The minimum number of specimen collected per sample units was five and the maximum 1,578, with five to 248 species. Of the total of 60 sample units, 11 had more than 70% of richness estimated by Jackknife 1, 15 sample units showed 65 to 70% and 14 sample units had richness between 60 and 65%. Twenty sample units had less than 60% of estimated richness, so that theyadded no new species to the whole checklist. The two CCA axes were not sufficient to explain the data, because only totalizing 19.5% of the data variation. The two DCA axes did not represent the spatial structure of the residuals based on Bonferronis sequential correction (axe 1: r²adj: 0,802; F: 26,348; p: <0,001; axe 2: r²adj: 0,547; F: 8,424; p <0,001). The best model for DCA axis 1 selected the following variables: precipitation of the hottest trimester, precipitation of the coldest trimester, mean temperature of the most humid trimester and the potential evapotranspiration to explain species composition. This model explains81.1% of the data variability. For axis 2, the variables altitude, precipitation of the coldest quarter and temperature seasonality were selected. Based on these results and on other studies carried out in Santa Catarina state, it can be concluded that conservation strategies have to be developed based on species distribution data and the environmental factors involved.