Browsing by Author "Bester, Cecile"
Now showing 1 - 1 of 1
Results Per Page
Sort Options
- ItemDetermining the phenotypic and molecular diversity within a collection of wheat wild relatives(Stellenbosch : Stellenbosch University, 2023-03) Bester, Cecile; Botes, Willem; Le Maitre, Nicholas; Stellenbosch University. Faculty of Agrisciences. Dept. of Genetics.ENGLISH ABSTRACT: Common bread wheat, Triticum aestivum L., is the third most important staple crop, after rice and maize. Wheat provides 20% of the globally consumed calories, and is cultivated on 242 389 108 ha globally. In South Africa, wheat was planted on 523 500 ha with a total production of 2 285 000 tonnes in the 2021/2022 season. The continuous improvement of wheat is important for global and local food security. Crop wild relatives are rich sources of genetic diversity that have been used in 4 157 documented cases of plant improvement by 2022. Wheat wild relatives, like Triticum and Aegilops spp. have been employed 333 times for wheat improvement by 2022. The Stellenbosch University Plant Breeding Laboratory has a collection of 1246 different wheat wild relatives belonging to the genera Triticum and Aegilops. There is limited information on the accessions from this collection, and the diversity present within, restricting the use of these plant genetic resources in wheat improvement. Obtaining more information that can assist when planning introductions, selections, renewal, description, and characterisation of the material is therefore vital. The development of digital phenotyping and characterisation methods can aid in the determination of morphological diversity and species identification. Chloroplast DNA is universal in plants and allows for the assessment of molecular diversity between species with different genomic combinations. From the Stellenbosch University Plant Breeding Laboratory crop wild relative collection, 92 entries were renewed, identified, characterised, and described. The collection showed a high level of morphological diversity in grain yield, flag leaf area and awn types. To improve phenotyping, a digital method of spike length estimation, using ImageJ, was compared to actual physical measurements. There was no significant difference between values obtained by these two methods, verifying the technique. Transfer Learning was applied to train pre-trained MobileNet Convolutional Neural Network models to distinguish between genus, species, subspecies and variety, with accuracies of 0.8750, 0.923, 0.742 and 0.600, respectively. These models can be applied to increase the accuracy of plant identification. Twelve chloroplast simple sequence repeats in four multiplex reactions were used to determine the genetic diversity within the wheat wild relative collection. Polymorphic information content ranged from 0.615 to 0.972 and gene diversity from 0.663 to 0.9724. These values indicate that there is a high level of diversity present within the wheat wild relative collection. Due to the high morphological, and molecular diversity within the ex situ species material collection, the plant genetic resources have a good potential for use in wheat breeding for crop improvement. The information can assist breeders to select the best wild relatives to be used in interspecies crosses to improve common bread wheat.