High Throughput, Single Kernel, Near Infrared Prediction and Sorting of Peanuts for Oleic Content

The discovery of high oleic peanut cultivars in the late 1980s offered peanut breeders the opportunity to develop a value added product with benefits to both commercial processors and consumers. Testing for this trait on single peanut kernels has traditionally been done via gas chromatography (GC), which is time consuming, expensive and requires removal of part of the seed, which can reduce seed germination. Near-infrared reflectance spectroscopy has been shown to successfully predict oleic content and linoleic content of both single kernels and bulk kernel lots. However, this process can still be time consuming and sorting of peanuts into separate high and normal oleic lots has to be done manually. In order for this technology to be successfully applied in breeding programs and commercial seed companies, a high throughput system for single kernel oleic prediction and sorting is required. Using the Brimrose Luminar 3076 Seedmeister NIR analyzer, absorbance spectra from 300 single kernels was collected from 1100 to 2300 nm. Several modifications had to be made to software and hardware components of the Seedmeister to enable accurate prediction and sorting, especially for larger Virginia size kernels. A partial least squares regression model was developed from the first derivative transformation of the raw spectra and 100 independent samples were then validated. This model successfully predicted and sorted between high oleic and normal oleic kernels based on oleic content. The model has been applied in the Australian Peanut Genetic Improvement Program to ensure high oleic purity in commercial seed lots.