A DYNAMIC SPATIAL MODEL OF SHIFTING CULTIVATION IN THE HIGHLANDS OF GUINEA, WEST-AFRICA

A dynamic spatial model of tropical deforestation and land-use change was developed from remotely sensed data for the Fouta Djallon mountain range in the Republic of Guinea, West Africa. The objective was to si mulate patterns of forest clearing for shifting cultivation in terms o f farmers' selection behavior for new fields based on topography and p roximity to villages. Data describing the current and historic conditi on of the vegetation cover and land use for a watershed in Guinea were derived from aerial photography and ground sampling. Maps of these co nditions were prepared and entered in a geographic information system (GIS) together with topographic data. From these data, maps of seconda ry variables (slope, village proximity, site productivity, and labor) were derived using the spatial operators contained in the GIS. These v ariables were ranked for agricultural preference and then combined fol lowing a pair-wise hierarchy to generate a composite agricultural site -preference map. Shifting agriculture was simulated by running the mod el through 18 iterations on a two-year time increment which correspond s to the typical duration of cultivation for any one field. An empiric ally derived number of cells from the agricultural site-preference map were converted from forest to active farms for each iteration to simu late deforestation. Different variable combinations and underlying ass umptions of model logic were tested to determine influence on simulati on results. To validate the model, the spatial characteristics of the projected landscape were compared with land-use data collected in 1989 , using tests for (1) size and distribution of agricultural sites, (2) image similarity (Kappa test), and (3) physical characteristics (slop e and distance from population centers) of the site. Although the mode l did not simulate the farmers' selection behavior for topography and village proximity successfully, test results with individual variables suggest that site productivity as determined by the length of fallow is a critical variable in the site selection process. Thus, aside from documenting the dynamics of shifting cultivation, this model allows p lanners to evaluate alternative strategies of land-use conversion with a graphic display of zones of potential hazards. Finally, the data co ntained in the GIS serve as a structure for monitoring long-term chang e in the region.