Figure 1.
Probability of species occurrence.
A) Anopheles dirus sensu lato, B) Anopheles dirus sensu stricto and C) Anopheles baimaii. Probability maps built using the mean of 100 replicates of the ECOOPT 1 km model based on 75% available samples. Inset maps present for each species presence/absence derived from the probability of species occurrence map based on 75% sample using as suitability threshold the value which maximizes sensitivity and specificity.
Figure 2.
Predicted potential and current distribution area for Anopheles dirus sensu lato.
The produced maps show in grey forested areas not suitable according to abiotic factor, in yellow the potential distribution based on abiotic factor but where forest is not present (potential niche) and the distribution as defined by favorable abiotic and biotic factors (“realized niche”). Performance tests for the model include test Gain (1.38), test AUC (0.90) and test extrinsic omission rate based on maximum test sensitivity plus specificity (6%).
Figure 3.
Predicted potential and current distribution area for Anopheles dirus sensu strico.
The produced maps show in grey forested areas not suitable according to abiotic factor, in yellow the potential distribution based on abiotic factor but where forest is not present (potential niche) and the distribution as defined by favorable abiotic and biotic factors (“realized niche”). Performance tests for the model include test Gain (2.091), test AUC (0.9554) and test extrinsic omission rate based on maximum test sensitivity plus specificity (2.5%).
Figure 4.
Predicted potential and current distribution area for Anopheles baimaii.
The produced maps show in grey forested areas not suitable according to abiotic factor, in yellow the potential distribution based on abiotic factor but where forest is not present (potential niche) and the distribution as defined by favorable abiotic and biotic factors (“realized” niche). Performance tests for the model include test Gain (1.7), test AUC (0.93) and test extrinsic omission rate based on maximum test sensitivity plus specificity (5.7%).
Figure 5.
Similarity dendrogram based on Ward clustering method and modified Hellinger distance.
Table 1.
Performance from univariate models.
Figure 6.
A) The maps depict the maximum distribution range based on extreme limit values. A graded shade of grey depicts areas with unfavorable value for one or more variables, white being favorable areas for A) Dirus complex, B) An. dirus s.s. and C) An. baimaii. Some sites represented by a star symbol were selected in unfavorable areas. D) A small table analyze if the value is higher (+) or lower (−) to the actual range of value for each variables.
Figure 7.
Comparison between Anopheles dirus sensu lato predicted distribution maps.
Illustration of differences between the A) predicted distribution developed using Boosted Regression Tree presented by the Malaria Atlas Project [8] and B) the current distribution map developed in this paper.