Persistency evaluation in a breeding context using UAV imagery

Perennial ryegrass (Lolium perenne L.) is one of the dominant forage grasses and is exploited in pastures for multiple years, thus an appropriate level of persistency is desirable to avoid extra costs due to early resowing, production losses, and nutritive value decline. Persistency is the capacity to survive and produce optimal yield over seasons. Therefore, it is an important evaluation criterion in breeding and is often scored after winter by visual observations made by experts, due to low cost and ease of implementation. But scoring is not completely independent of human bias.
Digital image analysis could remove the bias related to the observer. Persistency has been evaluated using imagery captured on-ground but the scale is limited to a plot. To solve this bottleneck we propose the use of Unmanned Aerial Vehicle (UAV) which allows to fly over the whole field and capture the plots simultaneously. UAV provides high spatial and temporal resolution imagery, quickly, at low cost and immediately accessible. Those are clear advantages compared to on-ground imagery.
The general objective was to develop a reliable methodology based on UAV imagery and derived Vegetation Indices (VIs) to estimate persistency of L. perenne accessions in a breeding context. Consistent VIs were found with a strong correlation with ground truth data. The degree of agreement between information of VIs generated using UAV-images and scoring for persistency was assessed.