 Bracts, the leaves subtending flowers, were lost multiple times in angiosperms, including in Brassicaceae, where their development is blocked early. Arabidopsis mutants that restore bract formation suggest that flower identity genes suppress the vegetative program of bract development, but the exact mechanisms and the evolutionary origin of bract loss remain unclear.

 We exploited natural variation in bracts that form only at the base of flowering branches in Arabidopsis, to study bract development and its connection to floral transition. We combined multiscale phenotyping, quantitative genetics, meristem imaging, time-series transcriptomics and curve registration to capture the genetic and developmental mechanisms unlocking bract development during floral transition.

 We mapped four Quantitative Trait Loci controlling bracts, excluding known bract mutant genes. We demonstrated LEAFY and other flower regulators were not involved and identified lists of new candidate genes and pathways, such as the anthocyanin pathway.

We found that bract develops when gene expression is desynchronised from the floral transition, either later or earlier, revealing a more complex landscape than the previously proposed prolonged vegetative state.

 We identified new mechanisms unlocking bract development. This natural variation sheds a new light on development canalisation during floral transition and on bract loss evolution.