Developing methods to map diversity in tropical forests is indispensable for the sustainable exploration, use, and conservation of these habitats. However, the lack of maps showing forest structure for large areas is a main restriction to building maps of diversity in tropical forest regions. We developed a methodology to map α-diversity of trees in tropical forest regions at high spatial resolution (50m) using α-diversity estimations of inventories as response variables and forest structural metrics and environmental variables as predictors. To include the forest structural metrics in our models, we first developed a method to map four of these metrics integrating LiDAR, multispectral, and SAR imagery. We evaluated these methods to map α-diversity of trees in the Chocó region of Colombia (South America), a lowland tropical moist forest with high tree diversity and complex forest structure.
Results/Conclusions
The relative errors (RE) of the Random Forest models used to map the four forest structural variables ranged from low to moderate; 11.7% for Vertical Canopy Heterogeneity, 14.3% for Canopy Height, 15% for standard deviation of normalized heights, and 17.4% for coefficient of variation of normalized heights. The maps of α-diversity also presented REs from low to moderate; the maps of Simpson and Shannon diversity indices obtained the lowest REs (6% and 13%), the map of richness had moderate RE (23%), and the maps of the effective number of species Shannon and Simpson presented the highest REs (47% and 64%). We found that the northeast and the southern side of the Chocó Region tended to have lower tree α-diversity than the central areas. The highest concentration of tree α-diversity is located along the pacific coast from the center to the northwest of the Chocó Region. We used open resources (software and imagery), except for the LiDAR data, in our analysis. In 2020 data from LiDAR GEDI (Global Ecosystem Dynamics Investigation lidar) will become available to any user. Thus, our methods can be applied to monitor the tree α-diversity of any tropical forest using GEDI data.