The Evolutionary History of Wild, Domesticated, and Feral Brassica oleracea (Brassicaceae)

Abstract

Understanding the evolutionary history of crops, including identifying wild relatives, helps to provide insight for conservation and crop breeding efforts. Cultivated Brassica oleracea has intrigued researchers for centuries due to its wide diversity in forms, which include cabbage, broccoli, cauliflower, kale, kohlrabi, and Brussels sprouts. Yet, the evolutionary history of this species remains understudied. With such different vegetables produced from a single species, B. oleracea is a model organism for understanding the power of artificial selection. Persistent challenges in the study of B. oleracea include conflicting hypotheses regarding domestication and the identity of the closest living wild relative. Using newly generated RNA-seq data for a diversity panel of 224 accessions, which represents 14 different B. oleracea crop types and nine potential wild progenitor species, we integrate phylogenetic and population genetic techniques with ecological niche modeling, archaeological, and literary evidence to examine relationships among cultivars and wild relatives to clarify the origin of this horticulturally important species. Our analyses point to the Aegean endemic B. cretica as the closest living relative of cultivated B. oleracea, supporting an origin of cultivation in the Eastern Mediterranean region. Additionally, we identify several feral lineages, suggesting that cultivated plants of this species can revert to a wild-like state with relative ease. By expanding our understanding of the evolutionary history in B. oleracea, these results contribute to a growing body of knowledge on crop domestication that will facilitate continued breeding efforts including adaptation to changing environmental conditions.

Keywords: Mediterranean; cabbage; crop wild relatives; domestication; ecological niche; origin.

Fig. 1.

Demographics and population structure for 224 samples of cultivated Brassica oleracea (n = 188) and wild C genome species (n = 36). (Left) Individual sample phylogeny with putatively wild samples labeled in bold and black dots indicating bootstrap values less than 70%. (Middle) Ancestry proportions for K = 2 to K = 5 as inferred from fastSTRUCTURE; K = 3 maximizes marginal likelihood (++) and K = 5 best explains structure in the data (+). (Right) Monophyletic clades indicated by a solid line, largest cluster of paraphyletic groups indicated by dashed lined. Illustrations of corresponding crop types by Andi Kur.

Fig. 2.

Principal component analysis (PCA) of SNPs and expression profiles. (A) Genetic variation PCA of PC1 versus PC2, (B) Genetic variation PC2 versus PC3, and (C) Expression profile PCA for PC1 versus PC2 of wild and cultivar samples. Triangles, wild samples; circles, cultivars; triangles with black outlines, WildC-2 samples with species identification indicated by color. Wild-collected B. cretica samples from Kioukis et al. (2020) indicated by asterisks, labeled as SRA.

Fig. 3.

Species tree with current distribution and historical environmental niche modeling. (A) Species tree of wild and cultivar samples. Bootstrap support indicated above branches. (B) Current species distribution of wild relatives. (C) Suitable habitat for B. cretica and (D) B. hilarionis during the late-Holocene. Map of current distrubution provided by Elizabeth Gjieli, the Geographical Information Manager at the New York Botanical Garden GIS Laboratory.

Fig. 4.

Inferred admixture events. (A) Phylogeny five migrations labeled a–f. (B) Corresponding four-population tests for treeness.