All the 18 partial warps that were extracted contributed significantly (α = 0.05) to the separation of the 10 groups (males and workers of the five species). Multivariate analysis of variance (MANOVA) demonstrated that the groups were significantly different (Wilk’s λ = 0.000001; P < 0.0001), and the cross validation test correctly classified 97.6% of the individuals to each respective group. The misidentifications were always between sexes in the same species and never among the species groups.
When analyzing only the males, all the 18 partial warps extracted contributed significantly (α = 0.05) to the discrimination of the five species and MANOVA demonstrated that the groups were significantly different (Wilk’s λ = 0.00002; P < 0.0001). The cross validation test was able to correctly identify 100% of the individuals in each respective group using the equations generated in the discriminant analysis. All individuals were classified with a probability of between 99.99 and 100% of belonging to each respective group. The graphical distribution of the groups (Figure 3) was similar to the one obtained analyzing males and females together. Again, Nannotrigona testaceicornis and Plebeia remota were placed closer to each other than to the other groups, which were more isolated from each other. The Mahalanobis square distances between the groups are presented in Table 2.