Habitat suitability modeling for orange (citrus) cultivation with an organic agriculture orientation under the impact of climate change in Phu Tho Province

Tóm tắt

Orange is a fruit crop of high economic value in the midland and mountainous region of Northern Viet Nam. In the context of climate change and the orientation toward organic agriculture, identifying areas suitable for orange cultivation is of major importance for sustainable production planning. This study applies the Random Forest (RF) algorithm within the biomod2 platform to assess and predict habitat suitability for orange cultivation in Phu Tho province under CMIP6 climate change scenarios.
The dataset comprises 14 variables in three groups: climate (WorldClim 2), soil-terrain (SoilGrids, DEM), and anthropogenic factors reflecting production development conditions. The model was trained using RF (ntree = 200, mtry = 2), evaluated by 5-fold cross-validation, and combined into an ensemble model (EMwmean). Future climate data were derived from the MPI-ESM1-2-HR model under two scenarios, SSP2-4.5 and SSP3-7.0, for the periods 2021-2040 and 2041-2060.
The results show that the model achieved high accuracy (AUC = 0.959; TSS = 0.886). Residential density, annual rainfall, and elevation were the factors with the greatest influence on habitat suitability. Under current conditions, 373,445.7 ha (39.89% of the natural area) were classified as suitable or highly suitable. Under the impact of climate change, the suitable area expands under both scenarios but to differing degrees: the expansion is larger and more stable under SSP2-4.5, whereas under SSP3-7.0 the gain is smaller in the early period (2021-2040) and the suitable area subsequently shifts toward higher-elevation areas. The findings provide a scientific basis for planning orange-growing areas oriented toward organic agriculture that are adapted to climate change.