Precision Phenotyping in Plant Breeding

Phenotyping large numbers of genotypes in field trials is central to achieve selection gain in plant breeding. However, field-based plant phenotyping has seen little innovations in recent decades. The utilization of sensor technology is a promising avenue to improve our ability to phenotype crops under field conditions. Here, I present the precision phenotyping platform ‘BreedVision’ and its utilization for plant phenotyping in combination with genomics approaches. The platform is equipped with several sensors, like for example light curtains, laser distance sensors, 3D-Time-of flight cameras and hyperspectral imaging, and enables a non-invasive high-throughput, high-dimensional phenotyping of crops under field conditions. The system architecture of the platform allows for sensor fusion, i.e. the simultaneous use of information from different, complementary types of sensors for trait prediction. The small grain cereal triticale (Triticosecale Wittmack) was used as a model crop and we performed a large calibration experiment for biomass yield. We obtained a high technical repeatability and prediction accuracy for biomass assessed at three different developmental stages. Using the established calibration models, biomass was predicted in a large mapping population with 647 doubled haploid triticale lines derived from 4 families. A multiple-line cross QTL mapping approach identified QTL that were common to all three different time points, but also developmental stage specific QTL. Taken together, the ‘BreedVision’ phenotyping platform is well suited for field-based phenotyping and the high-quality measurements enable to dissect the temporal changes of the genetic architecture underlying dynamic traits.