Integration and modularity in the male genitalia and parameres of Belostoma species of bifoveolatum group sensu Lauck, 1962 (Insecta, Heteroptera, Belostomatidae)

The bifoveolatum group comprises three known species (Belostoma angustum Lauck, B. bifoveolatum (Spinola), and B. elegans (Mayr)), with representatives occurring in southern and south-eastern South America, as well as along the Andes. These species, however, are cryptic with regard to body dimensions and male genitalia. Much of this cryptic variation in male genitalia is still misunderstood, so that proposed keys to bifoveolatum species are scarce and have been based only on body dimensions. In fact, it is not understood whether different parts of male genitalia covary themselves in the integrated way, or if parameres are somehow associated with male genitalia. This paper investigates whether there is expressive variation and integration in the shape of intromitent parts of male genitalia and parameres through specimens of those species from different southern South American localities, by using geometric morphometrics. We also test the hypotheses that the base of phallotheca [PHA]/dorsal arms [DA] subdivision, as well as base/apex of parameres, constitute separate modules. We semi- and landmarked PHA and DA, as well as the right paramere [PA], of 89 B. angustum males, four males of B. bifoveolatum and eight of B. elegans. To eliminate all non-shape variation, point configurations were subjected to the generalised least-squares Procrustes superimposition. A multilevel approach showed that the patterns of integration within individual, among individual and among populations centred by species are similar, with total variation concentrated in a few of the available dimensions. Accordingly, this correspondence is apparent with the very high values of matrix correlations between the covariance matrices for within-species-locality and among localities (r = 0.91 for PHA + DA and r = 0.86 for VD). Fluctuating asymmetry was not negligible for PHA + DA and it seems that there might be little agreement between it and the other levels of variation (matrix correlation using individual variation, r = 0.34. P < 0.0001; matrix correlation using variation among population, r = 0.29, P < 0.0001). After eliminating size effects, morphological integration between PHA + DA and PA was significant (RV = 0.058, P = 0.0313), which might have enabled greater elaboration of these parts resulting in their greater and faster diversification at the individual level. Our results indicated poor agreement between different levels of variation with regard to fluctuating asymmetry, such that direct developmental interaction is not important for all levels of covariation between PHA + DA and PA through bifoveolatum group of species. Our findings suggest the lack of diagnostic importance of the PHA + DA and VD to differentiate the species B. angustum and B. elegans, despite of their integration. Both DA/PHA and apex/base of PA modularity hypotheses were not refuted across all levels, which might be an indicative of the lack of integration between PA and PHA + DA at higher levels of organisation, such as among populations.