In the current proposal, we will develop new tools for effective potato (Solanum tuberosum) breeding. For targeted traits, we aim to i) develop fast and reliable high-throughput tools for common scab and potato tuber appearance phenotyping; ii) establish genomic selection as a routine tool in the Norwegian potato breeding program. Tuber exterior is extremely important for the consumers’ preferances and a cultivar’s market success. The project will target a disease that causes significant out-grading during the food value chain: the skin blemish disease common scab. In order to include this disease into the national potato selection program, a method for fast and reliable quantification needs to be developed. We propose to develop an automated high throughput platform for phenotyping this disease based on visual, near-infrared and thermal signals from an in-door studio set up. The signals will subsequently be analysed using deep learning models such as neural networks and machine learning. The intention is to build robust prediction models for phenotyping large number of samples based on image data. Genomic selection will be implemented in the national potato breeding program. Phenotypic data from the breeding material of the Norwegian potato breeder (Graminor Ltd.) will be collected and used to build genomic based prediction models for important tuber traits, including common scab, tuber size, tuber uniformity, tuber skin color, etc. The genomic marker data will be derived using the recently developed 22K GGP Potato SNP array. Various statistical approaches will be used for developing the genomic prediction model(s) including genomic best linear unbiased predictions and various Bayesian based approaches. The prediction model(s) will subsequently be validated before being applied to the breeding population for selection.
Project leader: Jahn Davik
Institution: NIBIO - NORSK INSTITUTT FOR BIOØKONOMI