Aerial insights into tree species diversity and biomass in a tropical agricultural landscape

Background/Question/Methods

At 38% of global land surface area, agriculture exceeds forest as the dominant global biome. Despite losses of continuous forest cover, tropical agricultural landscapes contain substantial dispersed tree cover in individual trees and tree clumps. Agricultural trees provide valuable services and products, such as seed sources, animal habitat, and carbon sequestration, however these are often not quantified due to inadequate methods to detect and characterize dispersed and heterogeneous tree cover. The objective of this research is to characterize the tree species composition and aboveground biomass of a tropical agricultural landscape in the Azuero Peninsula of Panama using high-resolution aerial hyperspectral and LiDAR (light detection and ranging) data from the Carnegie Airborne Observatory. First, we used hyperspectral reflectance (380-2510 nm) to develop a classification model to identify 25 tropical tree species. Second, we used LiDAR data (1.12 m spatial resolution) to estimate aboveground biomass of trees found in pastures and farms, in addition to estimating tree biomass of forested areas on the landscape. 

Results/Conclusions

We produce a classification model of 25 species that has an accuracy of 65.0% with species-specific accuracy ranging from 19.9-80.4%. While more than 25 tree species exist on the landscape, this work shows the potential for mapping species distributions across a large area in ways not feasible with ground surveys. Additionally, we estimate that approximately 54% of total landscape tree cover is found in pastures and fields, with an average aboveground biomass density of 12.7 Mg/ha. Across the 87 sq km landscape, the tree biomass found on farms accounts for more aboveground biomass than what is found in the riparian forests and forest fragments on the landscape. This work provides a way to scale-up the local knowledge of social, economic, and ecological importance of individual species to a landscape level. As agriculture continues to expand to provide resources for growing global populations, high resolution hyperspectral and lidar aerial data can provide information about dispersed trees in tropical agricultural landscapes.