In the early stages of plant breeding, breeders evaluate a large number of varieties. Due to limited availability of seeds and space, plot sizes may range from one to four rows. Spectral proximal sensors can be used in place of labour-intensive methods to estimate specific plant traits. The aim of this study was to test the performance of active and passive sensing to assess single and multiple rows in a breeding nursery. A field trial with single cultivars of winter barley and winter wheat with four plot designs (single-row, wide double-row, three rows, and four rows) was conducted. A GreenSeeker RT100 and a passive bi-directional spectrometer were used to assess biomass fresh and dry weight, as well as aboveground nitrogen content and uptake. Generally, spectral passive sensing and active sensing performed comparably in both crops. Spectral passive sensing was enhanced by the availability of optimized ratio vegetation indices, as well as by an optimized field of view and by reduced distance dependence. Further improvements of both sensors in detecting the performance of plants in single rows can likely be obtained by optimization of sensor positioning or orientation. The results suggest that even in early selection cycles, enhanced high-throughput phenotyping might be able to assess plant performance within plots comprising single or multiple rows. This method has significant potential for advanced breeding.
Keywords: border-row effect; high-throughput; phenomics; phenotyping; plant breeding; plot design; precision; spectral proximal sensing.