PMC

(A), Asian tiger mosquito (A. albopictus). (B), Western corn rootworm (D. v. virgifera).

Aa = A. albopictus, Dvv = D. v. virgifera, values above the red broken line are in the excellent band of the kappa index.

Pseudo-absence points plotted with presence points on the first three principal components of the training dataset (Species: D. v. virgifera), (A) SM1, (B) SM2, (C) SM3, and (D) SM4.

Error bars indicate standard errors. Bars with same letters are not significantly different (Tukey’s HSD test p>0.05), (A) specificity (B) sensitivity.

(A), SM1 pseudo-absences with model NNET (B) SM2 pseudo-absences with model KNN (C), SM3 pseudo-absences with model NNET (D) SM4 pseudo-absences with model SVM. Legend key: not suitable = p<0.4, low = 0.4<p<0.5, moderate = 0.5<p<0.7, high = 0.7<p<0.9, very high = p>0.9.

(A), SM1 pseudo-absences with model NNET (B) SM2 pseudo-absences with model NB (C), SM3 pseudo-absences with model CART (D) SM4 pseudo-absences with model KNN. Legend key: not suitable = p<0.4, low = 0.4<p<0.5, moderate = 0.5<p<0.7, high = 0.7<p<0.9, very high = p>0.9.

(A), SM1 pseudo-absences with model NNET (B) SM2 pseudo-absences with model NB (C), SM3 pseudo-absences with model CART (D) SM4 pseudo-absences with model KNN. Legend key: not suitable = p<0.4, low = 0.4<p<0.5, moderate = 0.5<p<0.7, high = 0.7<p<0.9, very high = p>0.9.

(A), SM1 pseudo-absences with model NNET (B) SM2 pseudo-absences with model KNN (C), SM3 pseudo-absences with model NNET (D) SM4 pseudo-absences with model SVM. Note: A. albopictus occurrence data is too dense to overlay on prediction map, refer to . Legend key: not suitable = p<0.4, low = 0.4<p<0.5, moderate = 0.5<p<0.7, high = 0.7<p<0.9, very high = p>0.9.

Error bars indicate standard errors over replicates. Bars with same letters within a graph are not significantly different (Tukey’s HSD test p>0.05). (A) model type, (B) pseudo-absence selection method, and, (C) species dataset.