Adult weight, wing size and shape of the four melanic morphs ofBactrocera dorsalis
After emergence, 30 males and 30 females of respective morphs were killed, individually weighed, and their right-wing excised from the thorax using a fine clamp. The insects were then preserved in in 70% ethanol. The removed wings were singly slide-mounted before image capture using a stereo microscope with LED and HD Camera Leica EZ4 HD (Leica Microsystems, Switzerland) at 16× magnification, to avoid deformation and enhance accuracy during photography and landmark collections.
The digital photographs were opened in the ImageJ software and Cartesian coordinates for 15 wing landmarks were generated (Fig. 1B). Using the same software, wing size parameters including (i) wing length (distance between the 5th and 15th landmark); (ii) wing width (distance between the 3rd and 13th landmark) (Fig. 1B); and (iii) wing area were generated. We used the PAST software V.3.09 (Hammer et al. 2001) to compute the centroid size parameter considered as a multidimensional measurement of wing size. It is calculated as the square root of the sum of squared Euclidean distances between each landmark and the wing centroid (Baleba et al. 2019). Also, based on the adult weight parameter, we calculated wing loading using the following formula:\(wing\ loading\ (\frac{\text{kg}}{m^{2})}=\frac{\text{mass}}{\text{wing\ area}}\ \).
To assess whether these wing size parameters were affected by the sex of the adults, the different level of melanin, and their interaction, we ran a two-way analysis of variance (ANOVA). The only significant effect that was conserved across the different wing size parameters was that related to the melanin level. For this, we only considered this factor in the 2 by 2 comparison, then performed an SNK posthoc test. All statistical comparisons were considered significant whenP  < 0.05. To identify correlations between centroid size, wing length, wing width, wing area, adult weight and wing loading, we performed separate principal component analysis (PCA) for four groups using the R packages called “FactoMineR” (Le et al. 2008) and “Factoextra” (Kassambara and Mundt 2020).
To determine the change of the wing shape across the different melanic morphs, we imported the raw landmark Cartesian coordinates into MorphoJ software (Klingenberg, 2011). We first performed a generalised Procrustes analysis to extract shape information from the data and eliminate differences in orientation, position and isometric size. Afterwards, using sex and level of melanin as factors, we executed a multivariate analysis of variance (MANOVA) to see whether these factors could impact the wing shapes of B. dorsalis . To visualise the wing shape deformations, we use the PAST software to generate the wing deformation grid of the wing of each sex and morph using the thin-plate spline analysis. Also, we used canonical variate analysis combined with discriminant analysis to analyse the relative similarities (or dissimilarities) of the different melanic morphs. To see the significance of pairwise differences in mean shapes, we performed permutation tests (10,000 rounds) with Procrustes distances.