Wing morphometric analysis using smartphone images and a free MATLAB-based tool: a case study with Apis mellifera from the Balearic Islands

Abstract

This study was designed to validate the new software WingMarks2 for the fully automatic and complete geometric morphometric analysis of the forewing venation structure in the honey bee and to apply this software for the morphometric analysis of Apis mellifera iberiensis in the Balearic Islands. Three experiments were performed to evaluate the ability of WingMarks2 software to analyze wing images obtained with different optical equipment (experiment 1), to check its ability to discriminate between six subspecies of Apis mellifera (experiment 2), to compare WingMarks2 and DeepWings software programs in their capabilities for detecting wing landmarks and associating them with specific subspecies (experiment 3) and to determine if there is a morphometric differentiation of A. m. iberiensis in the Balearic Islands when compared to those of the Iberian Peninsula (experiment 4). In experiment 1, WingMarks2 software allowed the automatic and accurate analysis of 91.7%, 95.2%, and 91.5% of the wings analyzed when using Leica microscope, Celestron USB microscope and Xiaomi Smartphone images, respectively. The results of experiment 2 showed that, overall, the cross-validated correct identification rate was 97.4% for the subspecies as a whole and ranged between 94% and 100% for the individual subspecies. In experiment 3, WingMarks2 proved more efficient than DeepWings in correctly processing the wings and assigning them to the subspecies. Finally, in experiment 4, the application of the new WingMarks2 software allowed us to identify a morphometric differentiation of A. m. iberiensis in the Balearic Islands when compared to those of the Iberian Peninsula.