Panicle size and architecture are traits of major relevance for oat production as they play a key role in conditioning the potential number of grains per spikelet and in consequence grain shape, size, and distribution. The aim of this research was to propose an image-based strategy and methodology to evaluate oat panicles. Specifically, the goals were to create and deploy the foundations for a high-throughput methodology to phenotype oat panicles and to determine the optimum number of panicles needed to detect differences between genotypic means for oat panicle traits with high power. A total of 48 genotypes were evaluated in four field experiments with three true replications in two locations and 2 years. Ten panicles per experimental unit were harvested and evaluated with image analysis using MATLAB. Panicle length, width, aspect ratio, compactness, and digital biomass were obtained from the images. The heritability of all panicle traits was high between 0.76 and 0.93. A high power of 0.80 to detect significant differences at an alpha of 0.05 among genotypic means for all panicle architecture traits was reached with the use of six panicles per replication. Our results showed a highly efficient and powerful strategy to study panicle architecture in oat. This methodology could be scaled-up to efficiently phenotype large populations and provide the basis for understanding panicle architecture and grain morphology in oat.

No abstract