Extended Data Fig. 1: Hierarchical cluster analysis of biomineralization signatures preserved in eggshell proteins (extant samples) and their fossilization products (fossil samples).

The topology represents a cluster analysis of n = 24 selected eggshell protein and PFP bands (Methods). Sampling of both biomineralized proteins (in situ analysis) from hard-shelled eggs and extracted, non-biomineralized membranes from soft and decalcified hard-shelled (Caiman, Alligator, Emys, Mesoclemmys, Phrynops and Gallus) eggs avoids phylogenetic attraction of the included fossil samples, and thereby allows eggshell clustering on the basis of the protein and PFP biomineralization signal. Two separate clusters of biomineralized and non-biomineralized eggshell proteins/PFPs are recovered. Pink nodes illustrate biomineralized egg proteins/PFPs, and blue nodes represent non-biomineralized eggshell proteins/PFPs. The egg icons illustrate whether samples represent originally hard or soft eggshell. One spectrum only was used for Mussaurus, as there is not much compositional variation across the eggshell (Fig. 1e), whereas all three eggshell spectra were sampled for Protoceratops, owing to the differences in composition across the egg section (Fig. 1d). Hard-shelled Alligator and turtle eggshells were excluded from this biomineralization analysis, as they do not produce any substantial organic signal with the spectroscopy protocol used (Supplementary Information and ref. ⁴⁴). Both Protoceratops and Mussaurus eggshells are nested within the cluster of originally non-biomineralized eggshell proteins/PFPs.