Ecosystem heterogeneity and lack of data on areas and deforestation rates h
ave constrained the modelling of deforestation in tropical open woodlands.
Smaller and more continuous changes in ecosystem quality attributes (e.g. c
anopy cover, tree density and biomass density) are best described as degrad
ation, not deforestation. Suggestions are made for constructing degradation
models by relaxing some assumptions of deforestation models to cope with t
he multiple uses of multiple vegetation layers by multiple users. Employing
biomass as a dependent variable would account for biomass reductions cause
d by changes in woodland area, tree density and the biomass of retained tre
es and shrubs, and overcome the problem of how to aggregate changes in wood
land area and tree density. Disaggregating models by woodland type is also
recommended.