Digital Phenotyping in Cotton Breeding Using Growth Rate Modeling Based on Visible Light Data Collected with Unmanned Aerial Systems

Abstract

The need to improve cotton to the benefit of humanity has been the driving factor in scientific cotton breeding for nearly 150 years. The act of cotton breeding itself is not a novel enterprise. What is novel, however, are the tools and technologies breeders utilize in cotton breeding operations. The semi-arid Texas High Plains agricultural area is one of the largest cotton producing regions in the world. This region is supported agriculturally by the ever-shrinking Ogallala aquifer. Identifying cotton varieties that perform or demonstrate yield stability across a range of water limited environments is critical to the long-term sustainability of cotton production on the Texas High Plains. Contemporary remote sensing devices like drones provide a pathway to evaluate cotton growth in a manner that is novel, efficient, and high throughput. The significance of this approach with respect to evaluation of cotton growth on the Texas High Plains is that information can be gathered over many breeding plots at high temporal granularity in a quantified and consistent manner. Observation of this type provides researchers with insight into growth patterns and characteristics not previously evident with manual plot-by-plot field observations. This presents the possibility that growth patterns and characteristics in breeding lines might be identified, when observed over a range of water levels in the semi-arid Texas High Plains, which provide a basis to isolate genotypes that will allow cotton production in this region to remain stable in the coming decades.