3.4.2. Models’ outputs and performance evaluation
The environmental predictive models displayed high predictive power according to the AUC values of 0.86, 0.93 and 0.84 for the species, Pop1 and Pop2, respectively (Table 3). These results suggest that the predictions effectively captured relationships between environmental variables and locations points of Pop1 comparing to Pop2 and the species.
Based on the Minimum training presence (MTP), suitable areas for Kersting’s groundnut and population groups were defined. We found that current distributions were significantly different between populations and species. The Maxent model for the species predicted a large area of cultivable conditions across the three agroclimatic zones, Southern Sudanian (SS), Northern Sudanian (NS) and Northern-Guinean (NG) (Fig. 6a1). The SS an NG zones were the areas forecasted to have high suitable climatic conditions for the species production. For the Pop1, the areas predicted to have high likely cultivability conditions were concentrated in NG zone of Benin, but very less and sparsely distributed in SS zone (Fig6b1). The Pop2 was projected across the three studied agroclimatic zones of the four countries with highest cultivable areas in SS and NS zones (Fig6c1).
Furthermore, the potential distribution maps under future (in 2055) climatic conditions revealed varied patterns in KG and genetic populations cultivable areas (Fig 6 and Fig 7, see supplemental Figures).
Under the two future climatic scenarios RCP4.5 and RCP8.5, an increase in the species cultivable areas for about 2.75%, were observed due to the decrease of the non-suitable areas (Fig 6a2, a3, Fig 7). This areas expansion was observed mainly in the NG zone of Southern Benin, also in SS and NS zones of Burkina Faso, Ghana and Togo. Similarly, an increase of the cultivability areas of Kersting’s groundnut was observed in the SS zone of the Northern Benin. On the other hand, the SS zone of Central Benin became climatically unsuitable to the crop production. The Pop1 showed to be more vulnerable to future scenarios as the suitable areas slightly decreased (0.504 % under RCP4.5 and 0.779 % under RCP8.5), while the unsuitable ranges increased (Fig 6b2, b3, Fig 7). The model of this genetic group predicted an increase in the suitable areas of NG zone of Benin, Ghana and Togo while a decrease was observed in the NS and SS zone of the four countries. The potential cultivable areas of the Pop2 (Fig 6c2, c3, Fig 7) slightly decreased by 0.322% under RCP4.5 while increased with extreme conditions of RCP8.5 (0.519%). The future climatic conditions of the SS zone of Central Benin would constitute constraints to this population cultivation. In contrast, the NG zone of Southern Benin and Togo and the SS zone gained in cultivable areas for Pop2.