Exploring UAV-imagery to support genotype selection in olive breeding programs

摘要

Airborne methodologies based on unmanned aerial vehicles (UAV) are becoming an extraordinary tool for implementing fast, accurate and affordable phenotyping strategies within plant breeding programs. The aim of this paper was to study the potential use of a previously developed UAV-OBIA platform, to fasten and support decision making for olive breeders regarding the selection of the most promising genotypes in terms of tree geometric traits. In particular, we have studied the feasibility of the system to efficiently classify and select olive genotypes according to four architectural parameters: tree height, crown diameter, projected crown area and canopy volume. These vegetative growth traits and their evolution during the first months after planting are key selection criteria in olive breeding programs. On-ground measurements and UAV estimations were recorded over two years (when trees were 15 and 27 months old, respectively) in two olive breeding trials using different training systems, namely intensive open vase and super high-density hedgerows. More than 1000 young trees belonging to 39 olive accessions, including new cross-bred genotypes and traditional cultivars, were assessed. Even though the accuracy in the UAV estimation compared to the on-ground measurements largely improved the second year, both methodologies detected in both years a high variability and significant differences among the studied genotypes, allowing for statistical comparisons among them. Genotype rankings based on the on-ground measures and UAV estimations were compared. The resulting Spearman's rank coefficient correlations were very high, at above 0.85 in most cases, which highlights that very similar genotype classifications were achieved from either field-measured or airborne-estimated data. Thus, UAV imagery may be used to assess geometric traits and to develop rankings for the efficient screening and selection of genotypes in olive breeding programs.