Fig. 1. Merianieae flower, landmark configuration and the five alternative hypotheses of floral modularity, visualized on an HRX-CT scan of Axinaea costaricensis (passerine syndrome).

Colour patterns represent the different hypothesized modules. Example of a fresh flower: important floral structures highlighted, sterile: pet—petal; male: app—stamen appendage; th—tubular anthers (thecae) containing pollen grains; po—stamen pore from where pollen is released; female: sty—style with stigma; only ethanol-preserved flowers were used in this study. 3D model landmarks: 37 landmarks placed on 3D-model of Merianieae flowers: 1–10—stamen appendage tips, 11–20—stamen appendage base, 21–30—stamen pores, 31—base of style, 32—stigma, 33–37—petal tips. Hypothesis 1: developmental modules—four organ whorls including the sterile petal whorl, the two stamen whorls (male organs; whorl 1 and 2), and the carpel whorl (female); the sepal whorl is not landmarked as it is not involved in pollination in Merianieae. Hypothesis 2: attraction module (showy, sterile petals) and reproduction module (male and female organs^12,13). Hypothesis 3: attraction module (showy petals and stamen appendages) and efficiency module (for pollen transfer, pore/stigma complex³¹). Hypothesis 4: alternative configuration of attraction module (colourful stamen appendages only) and efficiency module (petals, possibly also involved in mediating fit with the pollinator, and pore/stigma complex³¹). Hypothesis 5: Merianieae specific modules, attraction module (showy petals), pollen expulsion module (stamen appendages; function as handles for applying buzzes in the buzz-bee and as bellows organs for pollen expulsion in the passerine syndrome, but have lost their function in the mixed-vertebrate syndrome^25,28), and efficiency module (pore/stigma complex for pollen transfer).