The spatial and temporal distributions of vegetation can be influenced by edaphic and environmental factors as well as disturbances. The present study was undertaken to characterize the distribution and spatial dependence of species diversity in a fragment of semideciduous seasonal forest after a disturbance event by fire and to identify changes during natural regeneration. The fire, occurred in 1999, and surveys were undertaken in 2000, 2005 and 2014 in 26 geo-referenced plots (400 m²) distributed along two transects. The Fire Damage Index (FDI) in each plot was based on a scale from 0 to 5, and all of the arboreal individuals with diameters at breast height (DBH) ≥ 5 cm were identified in each plot. Species diversity was calculated using the Shannon index (H'), and species richness (S) was estimated using species accumulation curves; variations between the surveys compared based on the means and standard errors. The S, H’ and FDI data were submitted to non-spatial exploratory and geostatistical analyses. After adjusting the semivariograms, the levels of spatial dependence were classified, and interpolation of the variables values were performed using ordinary kriging to characterize their spatial distributions in the form of maps. Spatial analysis was used to identify and characterize differences between the post-fire surveys in terms of the configurations of the arboreal community. The fire event influenced the spatial and temporal structures of the variables S and H’. These variables showed spatial dependence and aggregated distributions, with reduction in the distance under spatial influences and a uniformity of individuals distribution in the forest fragment at the different surveys. This research characterized the distribution and spatial dependence of the variables S and H’ in a forest fragment after a fire event and the alterations in the arboreal community structure during natural regeneration. 

Keywords: Atlantic Forest, Brazil, forest fragmentation, anthropogenic disturbance, long-duration study, spatial dependence, geostatistics.