Modeling the distribution of the West Nile and Rift Valley Fever vector Culex pipiens in arid and semi-arid regions of the Middle East and North Africa

Table 3 Evaluation parameters for species distribution models excluding human population data

Model	Parameters in final model	#trees	Training AUC	CV AUC	Training COR	CV COR	CV deviance	Test COR (N = 158)	Test deviance (N = 158)
BRT Pop¹	9	1550	0.945	0.889	0.663	0.431	0.198	0.413	453.3
BRT No Pop²	14	3050	0.982	0.864	0.745	0.361	0.212	0.470	395.8

¹Data sources used for model development: Bioclim + Vegetation + Population (N = 24).
²Data sources used for model development: Bioclim + Vegetation (N = 23).
Evaluation parameters for Culex pipiens distribution models generated using boosted regression trees from a set of environmental predictors including human population density (“BRT Pop”) and a set excluding human population density (“BRT No Pop”). Models were built using 322 presence points, and evaluated using 10 fold cross validation. “CV AUC” and “CV Deviance” are the area under the curve (AUC) and Bernoulli deviance evaluated at the withheld dataset. “Test COR” and “Test Deviance” are calculated from the predicted model values at 79 independent occurrence points not used in training the mode, and 79 background points from the same region. “Test COR” = Pearson correlation coefficient.

ISSN: 1756-3305