The objective of this study is to better understand the complexity of defor
estation processes in southern Cameroon by testing a multivariate, spatial
model of land-cover change trajectories associated with deforestation. The
spatial model integrates a spectrum of independent variables that character
ize land rent on a spatially explicit basis. The use of a time series of hi
gh-spatial-resolution remote sensing images (Landsat MSS and SPOT XS), span
ning two decades, allows a thorough validation of spatial projections of fu
ture deforestation. Remote sensing observations reveal a continuous trend o
f forest clearing and forest degradation in southern regions of Cameroon, b
ut with a highly fluctuating rate. A significant proportion of the areas su
bject to a land-cover conversion experienced other changes in the following
years. The study also demonstrates that modeling land-cover change traject
ories over several observation years allows a better projection of areas wi
th a high probability of change in land-cover than projecting such areas on
the basis of observations from the previous time period alone. Statistical
results suggest that, in our southern Cameroon study area, roads mostly in
creased the accessibility of the forest for migrants rather than providing
incentives for a transformation of local subsistence agriculture into marke
t-oriented farming systems. The spacial model developed in this study allow
s simulations of likely impacts of human actions, leading to a transformati
on of the landscape (e.g., road projects) on key landscape attributes (e.g.
, biodiversity). Currently, several road projects or major logging concessi
ons exist in southern Cameroon.