Click here to add to your Google Calendar

Abstract:
Plant breeders spend most of their resources generating information on characteristics of breeding material that are relevant to growers and consumers of potential new varieties. This task, called phenotyping, is central to every breeding system, from traditional mass selection to marker-assisted selection to genomic selection and genetic engineering. In spite of the central role phenotyping plays in breeding programs, advances in phenotyping technology that lower cost and/or increase throughput have had most of their impact in only a few commodity crops. The majority of these technologies rely on light sensors including cameras and spectrophotometers paired with portable systems including smartphones and UAVs, enabling the rapid capture of unprecedentedly large amounts of data that can be translated to useful information for breeding decisions if coupled with proper data management and analysis. For specialty crop breeding programs, new phenotyping methods and analyses are often not readily applicable, meaning breeders continue to use traditional subjective rating approaches. While subjective ratings are effective for some traits, they have significant limitations due to error, throughput, and resolution that sensor-based phenotyping can overcome. Sweetpotato, Ipomoea batatas (L.) Lam. (2n=6x=90) is a globally significant food crop and a popular ornamental species, yet sweetpotato breeding programs lack sensor-based phenotyping approaches that work for its unusual and multidimensional traits. In this study, we adapt new phenotyping approaches to a wide range of traits that are important for tablestock and ornamental sweetpotato. First, we describe new methods for phenotyping ornamental sweetpotato (OSP) leaf shape and color based on smartphone imagery and portable spectrophotometry. We apply these methods to the NC State OSP diversity panel and demonstrate their utility in generating quantitative metrics of fit to desired product profiles based on shape and color, using image analysis and machine learning. We also investigate the molecular basis for the wide color spectrum observed in the OSP diversity panel using HPLC-DAD for anthocyanins and chlorophylls. For tablestock sweetpotato, we demonstrate the we demonstrate the use of UAV imagery for measuring storage root phenotypes, including a range of previously unavailable yield and shape parameters, using bedded sweetpotato storage roots as a proxy for freshly dug storage roots (due to the timing of the project), with significant time savings compared to traditional ground-based ratings and with similar accuracy for root counts and yield. Lastly, we present a suite of R Shiny web apps for breeding data management that fill a niche for highly modifiable, BrAPI-enabled tools that breeding programs can host in-house.