DATA - PREDICTING LANDSCAPE DYNAMICS AND NATIVE VEGETATION LOSS IN A COASTAL ISLAND BY MEANS OF SPATIALLY EXPLICIT MODELING

This repository contains datasets and outputs associated with the spatial modeling and simulation of land use and land cover (LULC) dynamics for the period 2012–2035 using the Dinamica EGO platform and the Bayesian Weights of Evidence (WoE) approach. The temporal scope of the historical analysis comprises the years 2012 and 2024, while simulations extend to future scenarios for 2030 and 2035. The year 2012 was adopted as the baseline because it marks the establishment of the Brazilian Forest Code (Law No. 12.651/2012), which legally recognizes mangroves and associated coastal dunes as Permanent Preservation Areas (APPs), providing legal protection against deforestation and other unauthorized anthropogenic disturbances. The repository includes: Transition probability maps (2013–2035) in GeoTIFF format generated using Dinamica EGO. These rasters represent spatial probabilities of LULC transitions estimated from static and dynamic explanatory variables. Probability values range from 0 (low transition likelihood) to 1 (high transition likelihood), while white or NaN areas indicate locations where a given transition is not applicable due to the absence of the source class. The rasters are projected in SIRGAS 2000 / UTM Zone 23S (EPSG:31983) with ~30 m spatial resolution. Each file contains 30 Float32 bands representing transitions among six LULC classes. Predictor variables folder, containing the explanatory variables used in the spatial modeling process. Eight predictor variables comprised the model: elevation; slop; soil organic carbon content; distance from urban areas; distance from rivers; distance from the coast; distance from mangrove areas; distance from roads The repository also includes annual LULC maps and datasets for 2012 and 2024 used for model calibration. LULC simulation outputs (2013–2035), containing annual simulated land use and land cover maps generated after model calibration and validation. Following validation procedures, the model simulated future scenarios for 2030 and 2035 while maintaining or adjusting transition rates and explanatory variables according to the study assumptions. These simulations allow the evaluation of potential spatial trajectories and future dynamics of land use and land cover. Weights of Evidence (WoE) outputs, including tables containing the WoE coefficients estimated for the explanatory variables used in the transition models. Each panel represents an environmental or anthropogenic variable, and the bars indicate WoE values associated with the discretized classes or intervals of each variable. The weights were estimated in Dinamica EGO using the Bayesian Weights of Evidence method based on the spatial relationship between observed LULC transitions and explanatory variables. Spatial dependence indices, Distribution of the values of the spatial dependence indices calculated for land use and land cover transitions. with Chi², Cramer, Contingency, Joint Entropy, and Joint Uncertainty values, highlighting the variability and statistical behavior of the association measures between the analyzed variables. Fuzzy Similarity Indices, for constant and exponential decay methods for the 2024 simulation, for Window 3x3 5x5 7x7 9x9 11x11 (Pixel X Pixel). To support the modeling process, the original LULC datasets were reclassified into six thematic classes adapted to the objectives of the study: (1) Mangrove – native mangrove vegetation, composed of trees and shrubs adapted to waterlogged and brackish soils, typical of tropical coastal areas; (2) Native dry soil vegetation – includes native tree vegetation formations that occur in environments not subject to flooding throughout the year, such as Forest Formation, Savanna Formation, and Herbaceous Restinga; (3) Native wet soil vegetation – includes native tree vegetation formations associated with environments subject to periodic or permanent flooding, including Floodplain Forest, Flooded Field, and swampy areas of the Apicum type; (4) Agriculture – areas intended for agriculture, farming, and livestock activity; (5) Walter bodies and sandy coastal environments – includes water bodies (rivers, lakes, and reservoirs), as well as sandy coastal environments, such as beaches, dunes, and exposed sand areas; and (6) Urban use – areas occupied by human settlements and associated infrastructure. These datasets support analyses of landscape dynamics, environmental change, spatial transition processes, and future land use scenario assessment in tropical coastal environments.