Tree cover distribution and above ground carbon stocks of dispersed trees in an agricultural landscape of the dry tropics

Background/Question/Methods

Deforestation in Central America in the mid-20th century turned once diverse tropical dry forests into landscapes dominated by pastures containing only fragments of the former forest. Though little forest remains, there is high potential for forest recovery because the network of dispersed trees and forest fragments encourage seed dispersal and provide suitable conditions for establishment. Conservation programs, such as REDD+, could play a major role in the recovery of these forests by providing payments to those who encourage reforestation; however, implementation is currently limited due to uncertainty in forest carbon estimates. Data from remote airborne or satellite platforms is becoming widely used to estimate and detect changes in forest cover and above ground carbon stocks, but despite major advances in this field, dispersed tree cover in agricultural landscapes is still not accounted for in current estimates. We use multiple endmember spectral mixture analysis, a sub-pixel image analysis technique, on 30 meter satellite imagery connected to high resolution images and field data to detect dispersed tree cover on the Azuero Peninsula of Panama. 

Results/Conclusions

We found that using a sub-pixel classification method of tree cover produced higher estimates of tree cover than traditional widely-used discrete classification methods. The amount of dispersed tree cover varied across the landscape, from no contribution to a significant increase tree cover in comparison to regional landcover estimates. These results can be used to more accurately estimate the above ground carbon stocks of agricultural landscapes, which is information necessary for the successful implementation of international conservation programs.