Phenotyping a diversity panel of quinoa using UAV-retrieved leaf area index, SPAD-based chlorophyll and a random forest approach

Abstract Given its high nutritional value and capacity to grow in harsh environments, quinoa has significant potential address a range of food security concerns. Monitoring the development phenotypic traits during field trials can provide insights into varieties best suited specific environmental conditions management strategies. Unmanned aerial vehicles (UAVs) promising means for phenotyping offer new relative plant performance. During trial exploring 141 accessions, UAV-based multispectral camera was deployed retrieve leaf area index (LAI) SPAD-based chlorophyll across 378 control saline-irrigated plots using random forest regression approach based on both individual spectral bands 25 different vegetation indices (VIs) derived from imagery. Results show that most VIs had stronger correlation with LAI measurements than bands. including red-edge band importance predictions, while near infrared (but not band) improved prediction models. When applied treatments (i.e. or saline), models trained all data saline data) achieved mapping accuracies (R 2 = 0.977–0.980, RMSE 0.119–0.167) 0.983–0.986, 2.535–2.861). Overall, study demonstrated remote sensing is only useful retrieving important quinoa, but machine learning available robust predictions abiotic stress experiments.