Background: Plant community assemblage is influenced by many factors, including soil characteristics and the arrival of diaspores from surrounding areas. These factors may be especially important in transition areas, leading to spatial gradients in the plant community. Methods: This study was performed in the understorey of an abandoned Eucalyptus grandis W. Hill ex Maiden plantation between a savanna and a forest, 490 m apart, in south-eastern Brazil. This study assessed whether the spatial variation in several variables related to the understorey's structure and composition is best described by linear or non-linear (quadratic) models. The linear model would indicate a gradient between the two vegetation types, whereas the quadratic model would indicate a stronger effect of the plantation's edges. Results: There was a gradient in species composition between the two edges of the plantation. Mean vegetation height was greatest at the savanna edge and lowest in the centre of the plantation. The total number of individuals per plot and the phylogenetic diversity decreased with distance from the savanna edge. Different patterns were observed for different dispersal syndromes, with animal dispersal being more common at the savanna edge, wind dispersal in the centre of the plantation and self dispersal at the forest edge. Conclusions: The greater number of individuals at the savanna edge may indicate that dispersal and arrival of diaspores are the most important factors influencing community structure and composition of the understorey of this abandoned E. grandis plantation, with most propagules coming from the savanna area. The smaller vegetation height in the centre of the fragment may also indicate older colonisation at the edges. Therefore, in addition to highlighting the recovery potential of undergrowth beneath abandoned Eucalyptus spp. plantations, these results show that this recovery is spatially heterogeneous and that dispersal plays a large role in it. This should be taken into account in restoration projects. The authors recommend careful consideration before removing regenerating Eucalyptus spp. trees as part of the site restoration. Instead the focus should be on the recovery potential of the undergrowth, with gradual removal of Eucalyptus trees, if necessary.

Originality measures how different a given species is from all other co-occurring species regarding either their phylogenetic history or functional traits. Since it is important to preserve the various aspects of diversity and original species carry more phylogenetic or functional information, originality may be used to assign conservation priorities. Our goal was to evaluate the relationships between phylogenetic and functional originalities, and their simulated losses under extinction scenarios based on abundance, fire tolerance and habitat preference. We placed 100 plots in a cerrado reserve located in central Brazil, sampled all woody plants species within the plots, measured 14 functional traits and measured fire history. We assembled a phylogenetic tree and a functional dendrogram, with which we calculated the originalities. Phylogenetic-and functional-based originalities were correlated. However, the loss of functional originality was different from random extinctions on the abundance and fire tolerance scenarios, whereas the loss of phylogenetic originality was not. When compared with phylogenetic originality, functional originality brought more information to be used in conservation strategies because it was sensitive to differences in species abundance and fire tolerance. Thus, the extinction of rare or fire-sensitive species would result in important functional changes due to loss of distinctive traits.