Petals can be simple or elaborate, depending on whether they have lobes, teeth, fringes, or appendages along their margins, or possess spurs, scales, or other types of modifications on their adaxial/abaxial side, or both. Elaborate petals have been recorded in 23 orders of angiosperms and are generally believed to have played key roles in the adaptive evolution of corresponding lineages. The mechanisms underlying the formation of elaborate petals, however, are largely unclear. Here, by performing extensive transcriptomic and functional studies on Nigella damascena (Ranunculaceae), we explore the mechanisms underlying elaborate petal development and specialized character formation. In addition to the identification of genes and programs that are specifically/preferentially expressed in petals, we found genes and programs that are required for elaborate rather than simple petal development. By correlating the changes in gene expression with those in petal development, we identified 30 genes that are responsible for the marginal/ventral elaboration of petals and the initiation of several highly specialized morphological characters (e.g., pseudonectaries, long hairs, and short trichomes). Expression and functional analyses further confirmed that a class I homeodomain-leucine zipper family transcription factor gene, Nigella damascena LATE MERISTEM IDENTITY1 (NidaLMI1), plays important roles in the development of short trichomes and bifurcation of the lower lip. Our results not only provide the first portrait of elaborate petal development but also pave the way to understanding the mechanisms underlying lateral organ diversification in plants.

AP1/FUL, SEP, AGL6, and FLC subfamily genes play important roles in flower development. The phylogenetic relationships among them, however, have been controversial, which impedes our understanding of the origin and functional divergence of these genes. One possible reason for the controversy may be the problems caused by changes in the exon-intron structure of genes, which, according to recent studies, may generate non-homologous sites and hamper the homology-based sequence alignment. In this study, we first performed exon-by-exon alignments of these and three outgroup subfamilies (SOC1, AG, and STK). Phylogenetic trees reconstructed based on these matrices show improved resolution and better congruence with species phylogeny. In the context of these phylogenies, we traced evolutionary changes of exon-intron structures in each subfamily. We found that structural changes have occurred frequently following gene duplication and speciation events. Notably, exons 7 and 8 (if present) suffered more structural changes than others. With the knowledge of exon-intron structural changes, we generated more reasonable alignments containing all the focal subfamilies. The resulting trees showed that the SEP subfamily is sister to the monophyletic group formed by AP1/FUL and FLC subfamily genes and that the AGL6 subfamily forms a sister group to the three abovementioned subfamilies. Based on this topology, we inferred the evolutionary history of exon-intron structural changes among different subfamilies. Particularly, we found that the eighth exon originated before the divergence of AP1/FUL, FLC, SEP, and AGL6 subfamilies and degenerated in the ancestral FLC-like gene. These results provide new insights into the origin and evolution of the AP1/FUL, FLC, SEP, and AGL6 subfamilies.