Differentiating vegetation types from eastern South American ecosystems based on modern and subfossil pollen samples: evaluating modern analogues

Abstract

In south and southeast Brazil land use caused profound changes in natural vegetation and consequently the value of the pollen composition in surface samples as modern analogues. In order to test the capability of modern pollen to represent the natural vegetation, three different time slices of pollen assemblages from 27 sites spread over southern and south-eastern Brazil and the Misiones Province in Argentina were collated. Pollen samples from the pre-colonization period, selected from the moment just before abrupt changes evidenced on pollen diagrams caused by the colonization process throughout the last 500 years, were assumed to represent the natural vegetation conditions once the climate remained stable within this period. Thus we used pre-colonization assemblages to compare with modern samples to explore to what extent surface pollen may be biased in representing the natural vegetation types. Furthermore, to compare man made vegetation change to climate driven vegetation change we also compared to these 20 out of 27 samples dated to 3,000 years bp. Guided by ordination and cluster analysis, but using abundance thresholds of indicator taxa we classified the pollen spectra of pre-colonization time into seven groups consistent with the main vegetation types in the area. Ordination analyses capture the differentiation between grassland and forested vegetation and between tropical and subtropical vegetation types. Comparing the pre-colonization with other time slices we observed that based on Poaceae abundance, 70 and 85 % respectively of sites from 3,000 bp and modern assemblages maintained their classification. Based on finer classification criteria these values decreased to 40 and 52 % respectively. Square chord dissimilarity indicates that colonization impact altered the pollen composition as strongly as 3,000 years of climate induced vegetation change. The surface samples still represent important environmental gradients; however, their use as modern analogue requires careful treatment and eventual exclusion of highly impacted sites.