Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Accueil du site → Doctorat → États-Unis → 2019 → Implementing Unmanned Aerial Systems Within a Field-Based Maize (Zea mays L.) Breeding Program : Improving Yield Prediction and Understanding Temporal QTL Expression of Plant Height

Texas A&M University (2019)

Implementing Unmanned Aerial Systems Within a Field-Based Maize (Zea mays L.) Breeding Program : Improving Yield Prediction and Understanding Temporal QTL Expression of Plant Height

Anderson II, Steven Langlie

Titre : Implementing Unmanned Aerial Systems Within a Field-Based Maize (Zea mays L.) Breeding Program : Improving Yield Prediction and Understanding Temporal QTL Expression of Plant Height

Auteur : Anderson II, Steven Langlie

Université de soutenance : Texas A&M University

Grade : Doctor of Philosophy (PhD) 2019

Résumé
Unmanned aerial system (UAS) technologies are becoming common place within field-based agriculture programs allowing breeders to evaluate greater numbers of genotypes, reducing resource inputs and maintaining unbiased data collection. A comprehensive evaluation was conducted focused on the implementation of UAS technologies within a field-based maize breeding program using the plant height phenotype as a proof of concept in implementation and validation. A robust data processing pipeline was developed to extract height measurements from RGB structure from motion (SfM) point clouds. The 95th percentile (P95) height estimates exceeded 70% correlation to manual ground truth measurements across diverse germplasm groups of hybrid (F1) and inbred lines. Sigmoidal functions were developed to model the overall growth and trajectory of hybrids (R^2 : >98% ; RMSE : < 14 cm) and inbred (R^2 : >99% ; RMSE : < 4 cm). UAS-based height estimates demonstrated greater capacity to partition phenotypic variance to genetic components compared to manual measurements ; function growth parameters (asymptote, inflection point, and growth rate) were explained by more than 70% of variance with genetics for the hybrid trials. UAS height estimates improved correlations to hybrid grain yield >1.5- fold similar to functional growth parameters. A 4-fold improvement in indirect selection of hybrid grain yield was achieved using functional growth parameters compared to conventional manual, terminal plant height (PHTTRML). We expanded our implementation of UAS phenotyping to evaluate three inbred line mapping populations aimed at studying functional QTL and temporal QTL expression. Functional growth parameters identified 34 associations explaining 3 to 15% genetic variation. Height was estimated at one-day intervals to 85 DAS using the Weibull function, identifying 58 unique temporal peak QTL locations. Temporal QTL demonstrated all of the identified significant QTL had dynamic expression patterns. In all, UAS technologies improved phenotypic selection accuracy and have capacity to monitor traits on a temporal scale furthering our understanding of crop development and biological trajectories.

Présentation

Page publiée le 26 mai 2021