John Libbey Eurotext




Spatial prediction and distribution modelling of Acacia spp. in Eastern Burkina Faso
Science et changements planétaires / Sécheresse. Volume 19, Number 4, 283-92, octobre-novembre-décembre 2008, Article de recherche
Résumé Article gratuit
Author(s) : Salifou Traoré, Oumar Kaboré, Lamourdia Thiombiano, Jeanne Rasolodimby-Millogo
Summary : Spatial prediction is an important tool for ecosystem planning and biodiversity conservation. With regard to socio-economic and ecological issues of Acacia species for rural development of drylands (forage for livestock, fuel wood, gum production, restoration of degraded land), the distribution of four Acacia spp. (Acacia dudgeoni, Acacia gourmaensis, Acacia hockii and Acacia seyal) was modelled in a sudano-sahelian zone (eastern Burkina Faso). This modelling is based on a sampling of 175 plots fitting environmental (climate, soil) and geographic variables (local density, spatial trend) as the predictors of species occurrence. The models are selected and evaluated with cross validation methods using GRASP (Generalized Regression and Spatial Predictions). Validation showed that the models of A. gourmaensis and A. hockii are stable and adequate with high values of Spearman correlation between observed and predicted distribution by cross validation (respectively 0.72 and 0.84). The model of A. seyal shows a significant spatial autocorrelation and it spatial prediction is calibrated by two methods considering autoregressive terms. Spatial prediction using spatial coverage of predictors showed the potential distribution of Acacia spp., highlighting the areas of high and low occurrence. Such information is useful for the management and valorisation of these forest resources under climate change and growing human pressure.
Keywords : cross validation, predictor, regression model, spatial coverage

Spatial prediction and distribution modelling of Acacia spp. in Eastern Burkina Faso