Different Shades of Kale—Approaches to Analyze Kale Variety Interrelations

Abstract

Brassica oleracea is a vegetable crop with an amazing morphological diversity. Among the various crops derived from B. oleracea, kale has been in the spotlight globally due to its various health-benefitting compounds and many different varieties. Knowledge of the existing genetic diversity is essential for the improved breeding of kale. Here, we analyze the interrelationships, population structures, and genetic diversity of 72 kale and cabbage varieties by extending our previous diversity analysis and evaluating the use of summed potential lengths of shared haplotypes (SPLoSH) as a new method for such analyses. To this end, we made use of the high-density Brassica 60K SNP array, analyzed SNPs included in an available Brassica genetic map, and used these resources to generate and evaluate the information from SPLoSH data. With our results we could consistently differentiate four groups of kale across all analyses: the curly kale varieties, Italian, American, and Russian varieties, as well as wild and cultivated types. The best results were achieved by using SPLoSH information, thus validating the use of this information in improving analyses of interrelations in kale. In conclusion, our definition of kale includes the curly varieties as the kales in a strict sense, regardless of their origin. These results contribute to a better understanding of the huge diversity of kale and its interrelations.

Glucosinolates and Cytotoxic Activity of Collard Volatiles Obtained Using Microwave-Assisted Extraction

“…var. medullosa L.) [ 3 , 4 ]. Kale’s younger, more delicate leaves are used for human consumption, whilst its older leaves are better suited for use as animal fodder [ 5 ].…”

“…Considering the genetic diversity between 72 different varieties of kale and cabbage, recent research making use of the single nucleotide polymorphism approach suggests that non-curled American kales should be correctly treated as collards (var. viridis ) and not considered a kale group [ 4 ].…”

A Cold Case—Glucosinolate Levels in Kale Cultivars Are Differently Influenced by Cold Temperatures

“…Representatives of the three main kale groups [2] were chosen, i.e., curly-type kale "Frostara" (obtained from Bruno Nebelung GmbH, Everswinkel, Germany), Lacinatotype kale "Palmizio" (obtained from Thompson & Morgan Ltd., Ipswich, UK), and wild cabbage "Helgoländer" (feral type, seeds harvested at the Botanical Garden Oldenburg) as representative for collard kale types (Figure 1). 45 individuals of each cultivarwere seeded in multipot trays (substrate Hawita-Flor P + Ton 4007, HAWITA Gruppe GmbH, Vechta, Germany) and cultivated in a climate chamber (const.…”

“…The general versatility of kale is paralleled by an enormous range of plants denominated as "kale", comprising more than 120 different varieties and cultivars, of which some are only locally known and which occur from the Mediterranean climate to a north temperate climate. This diversity can be grouped as either "curly kale", "Lacinato kale", "collards", "Russian/Siberian kale", or "feral types", as demonstrated by Hahn et al [2]. They differ not only genetically, but also in terms of growth, leaf morphology, color, phytochemicals, and taste [3].…”

A Study of Kale Recognition Based on Semantic Segmentation

“…In this context, the following limitations of existing research are summarized by this study: (1) insufficient datasets: the lack of comprehensive datasets capable of capturing kale diversity under real-world conditions hinders the training and validation of robust segmentation models [15]; (2) limited feature representation: existing methods often cannot effectively represent and extract the unique features of kale bulbs and outer leaves, especially in scenes with complex backgrounds [11]; (3) edge detection accuracy: the accuracy of edge detection of kale leaves often falls short of the requirements, which are crucial for accurate segmentation [10]; (4) class imbalance: the class imbalance problem during training has not been fully resolved, which may lead to biased model performance [14].…”