Development and Implementation of a Small Unmanned Aerial Systems Based Phenotyping Pipeline for Plant Breeding Programs.

The combination of robust, scalable phenotyping tools and genotyping tools can facilitate the rapid genetic gains in breeding programs.  While the genotyping technologies have made significant progress in post-human genome sequencing era, advances in phenotyping technologies have been slower to develop. Now, emerging next generation sensing and robotic technologies are set to transform the field of high-throughput phenotyping (HTP).  A Small Unmanned Aerial System or sUAS, fitted with lightweight, spectral imaging sensors, enables high-throughput, precision analysis of a large number of breeding plots. Following the semi-automatic image processing and data analysis pipeline developed by the Poland Lab at Kansas State University, approximately 28,500 and 41,900 plot level Green-NDVI and Digital Elevation Model (DEM)-derived plant height measurements, collected in India in years 2015 and 2016 are currently being processed.  We evaluated the predictive ability of proximal measurements to target traits of biomass and grain yield.  Across multiple trials and dates, we found significant correlations between Green-NDVI and biomass (r = 0.60-65***), and Green-NDVI and grain yield (r = 0.41-0.58***) over the course of the growing season.  The within-site broad-sense heritability (H²) ranged from 0.60 to 0.88 for DEM-derived plant height.  The relationship of Green-NDVI and plant height with other physiological and agronomic traits was also analyzed.  will also be explored. This initial large-scale deployment has validated and improved the data management protocols, image processing pipelines and baseline validation of the measurements from the sUAS.  While the preliminary findings demonstrate a successful implementation of sUAS based HTP pipeline, the future work will focus on optimization of the downstream data extraction routines, and to explore alternate G-by-E and multi-trait modeling schemes.