Fig. 2. Trees obtained from the equal weights parsimony analyses with some characters deactivated; only the corbiculate bees are shown. (a) Strict consensus of six most parsimonious trees from the analysis with all mouthpart characters deactivated (53 chars deactivated; strict consensus: L = 954, CI = 0.36, RI = 0.79). (b) Strict consensus of 32 most parsimonious trees from the analysis with all genitalia/sting characters deactivated (57 chars deactivated; strict consensus: L = 961, CI = 0.34, RI = 0.79). (c) Strict consensus of 17 most parsimonious trees from the analysis with all mouthpart and genitalia/sting characters deactivated (110 chars deactivated; strict consensus: L = 795, CI = 0.32, RI = 0.78). (d) Strict consensus of 98 most parsimonious trees from the analysis with all characters possibly under some morpho-functional constraint and/or potential selective pressures deactivated (138 chars deactivated, see the Methods section for details; strict consensus: L = 690, CI = 0.30, RI = 0.75). Photographs of A. dorsata, B. pauloensis, C. collaris, and M. quadrifasciata taken from Porto et al. (2021).