Fruits exhibit a vast array of different 3D shapes, from simple spheres and cylinders to more complex curved forms; however, the mechanism by which growth is oriented and coordinated to generate this diversity of forms is unclear. Here, we compare the growth patterns and orientations for two very different fruit shapes in the Brassicaceae: the heart-shaped Capsella rubella silicle and the near-cylindrical Arabidopsis thaliana silique. We show, through a combination of clonal and morphological analyses, that the different shapes involve different patterns of anisotropic growth during three phases. These experimental data can be accounted for by a tissue-level model in which specified growth rates vary in space and time and are oriented by a proximodistal polarity field. The resulting tissue conflicts lead to deformation of the tissue as it grows. The model allows us to identify tissue-specific and temporally specific activities required to obtain the individual shapes. One such activity may be provided by the valve-identity gene FRUITFULL, which we show through comparative mutant analysis to modulate fruit shape during post-fertilisation growth of both species. Simple modulations of the model presented here can also broadly account for the variety of shapes in other Brassicaceae species, thus providing a simplified framework for fruit development and shape diversity.

SummaryEvolution of gene-regulatory sequences is considered the primary driver of morphological variation [1, 2, 3]. In animals, the diversity of body plans between distantly related phyla is due to the differential expression patterns of conserved “toolkit” genes [4]. In plants, variation in expression domains similarly underlie most of the reported diversity of organ shape both in natural evolution and in the domestication of crops [5, 6, 7, 8, 9]. The heart-shaped fruit from members of the Capsella genus is a morphological novelty that has evolved after Capsella diverged from Arabidopsis ∼8 mya [10]. Comparative studies of fruit growth in Capsella and Arabidopsis revealed that the difference in shape is caused by local control of anisotropic growth [11]. Here, we show that sequence variation in regulatory domains of the fruit-tissue identity gene, INDEHISCENT (IND), is responsible for expansion of its expression domain in the heart-shaped fruits from Capsella rubella. We demonstrate that expression of this CrIND gene in the apical part of the valves in Capsella contributes to the heart-shaped appearance. While studies on morphological diversity have revealed the importance of cis-regulatory sequence evolution, few examples exist where the downstream effects of such variation have been characterized in detail. We describe here how CrIND exerts its function on Capsella fruit shape by binding sequence elements of auxin biosynthesis genes to activate their expression and ensure auxin accumulation into highly localized maxima in the fruit valves. Thus, our data provide a direct link between changes in expression pattern and altered hormone homeostasis in the evolution of morphological novelty.

The transition from pollinator-mediated outbreeding to selfing has occurred many times in angiosperms. This is generally accompanied by a reduction in traits attracting pollinators, including reduced emission of floral scent. In Capsella, emission of benzaldehyde as a main component of floral scent has been lost in selfing C. rubella by mutation of cinnamate-CoA ligase CNL1. However, the biochemical basis and evolutionary history of this loss remain unknown, as does the reason for the absence of benzaldehyde emission in the independently derived selfer Capsella orientalis.We used plant transformation, in vitro enzyme assays, population genetics and quantitative genetics to address these questions.CNL1 has been inactivated twice independently by point mutations in C. rubella, causing a loss of enzymatic activity. Both inactive haplotypes are found within and outside of Greece, the centre of origin of C. rubella, indicating that they arose before its geographical spread. By contrast, the loss of benzaldehyde emission in C. orientalis is not due to an inactivating mutation in CNL1.CNL1 represents a hotspot for mutations that eliminate benzaldehyde emission, potentially reflecting the limited pleiotropy and large effect of its inactivation. Nevertheless, even closely related species have followed different evolutionary routes in reducing floral scent.

The study of the independent evolution of similar characters can highlight important ecological and genetic factors that drive phenotypic evolution. The transition from reproduction by outcrossing to self-fertilization has occurred frequently throughout plant evolution. A common trend in this transition is the reduction of flower features in the selfing lineages, including display size, flower signals and pollinators' rewards. These changes are believed to evolve because resources invested in building attractive flowers are reallocated to other fitness functions as the pressures to attract pollinators decrease. We investigated the similarities in the evolution of flower fragrance after independent transitions to self-fertilization in Capsella. We identified a large number of compounds that are similarly changed in different selfer lineages, such that the composition of the flower scent can predict the mating system in this genus. We further demonstrate that the emission of some of these compounds convergently evolved based on mutations in different genes. In one of the Capsella selfing lineages, the loss of β-ocimene emission was caused by a mutation altering subcellular localization of the ortholog of TERPENE SYNTHASE 2 without apparent effects on its biosynthetic activity. This mutation appears to have been selected at the early stage of this selfing lineage establishment through the capture of a variant segregating in the ancestral outcrossing population. The large extent of convergence in the independent evolution of flower scent, together with the evolutionary history and molecular consequences of a causal mutation, suggest that the emission of specific volatiles has important fitness consequences in self-fertilizing plants without obvious energetic benefits.