Virtual phenomics - use of robots and drones in combination with genomics accelerate genetic gains in wheat breeding
Files
Date
2019-10-22
DOI
Authors
Shafiee, Sahameh
From, Pål
Burud, Ingunn
Dieseth, Jon Arne
Vindfallet, Are
Crossa, Jose
Alsheikh, Muath
Version
Supporting documentation
OA Version
Citation
Sahameh Shafiee, Pål From, Ingunn Burud, Jon Arne Dieseth, Are Vindfallet, Jose Crossa, Muath Alsheikh. 2019. "Virtual phenomics - use of robots and drones in combination with genomics accelerate genetic gains in wheat breeding." 6th International Plant Phenotyping Symposium. Nanjing, China, 2019-10-22 - 2019-10-25.
Abstract
Wheat breeding is a tedious process that usually takes 10-15 years and depends heavily on the ability to identify superior progeny lines by visual inspection and manual scoring of traits. Two emerging technologies are now offering potential for more precise selection and faster genetic gains: genomic prediction of breeding values based on genome-wide SNP markers and use of high throughput phenotyping technologies. In the innovation project “Reliable and efficient high-throughput phenotyping to accelerate genetic gains in Norwegian plant breeding (virtual phenomics; vPheno), 2017-2022” we are combining multispectral imaging with genomic prediction. This is a collaborative project between the industry partners Graminor AS and Making View AS and world-leading research groups in genetics, robotics and image analysis at the Norwegian University of Life Sciences, Boston University and the International Maize and Wheat Improvement Center (CIMMYT) in Mexico. In order to follow the growth of the plants during the season and calculating vegetation indices that can be used to predict grain yield, the project makes use of drones fitted with multispectral camera that are flown at weekly interval during the field season. In addition, a custom-built field robot is being used for gathering close-up images of field plots that will be used for counting the number of heads per square meter and other plant features that cannot be reliably recognized from drone images. One major use of the data is to improve the precision of genomic prediction models, the other is to enable plant breeders to visit field trials in "virtual reality", by integrating information from the drone and robot images with other available data on the field plots (grain yield, disease resistance, quality traits, marker data etc.). A prototype of the VR tool will be presented along with the progress on improving grain yield prediction by use of the multispectral drone images.