Fig 1.
Average daily distribution of temperature and precipitation for trials carried out in the summer of 2014 (temperature: solid red line; precipitation: black bars) and 2016 (temperature: dashed blue line; precipitation: green bars) under well-watered and drought stressed conditions in Ciudad Obregon (Sonora, Mexico) relative to the planting date.
Arrows indicate date of five flights starting at flowering (F1 to F5). Flights took place 55 (F1), 62 (F2), 69 (F3), 75 (F4) and 83 (F5) days after planting.
Fig 2.
Distribution of trait values and repeatability for trials carried out under well-watered (blue boxplots) and drought stressed (red boxplots) conditions for trials carried out in 2014 and 2016.
Phenotypic traits shown are grain yield (A), days to anthesis (B), plant height (C) and the anthesis silking interval (ASI; D).
Fig 3.
Reflectance (left column) and repeatability for individual bandwidths (right column) across the spectral range from 392 nm to 850 nm for individual flights in the well-watered (blue line) and drought stressed (red line) treatment in 2014 (solid line) and 2016 (dotted line).
Fig 4.
Prediction accuracy for within and across environment predictions using different statistical methods and input parameters.
Statistical methods used were partial least square regression (PLSR), random forest (RF), ridge regression (RR) and Bayesian ridge regression (BayesB). Input factors used for model parametrization were molecular markers (M), hyperspectral reflectance data (H) and the combination of both (HM).