Background: The seedling stage has received little attention in maize breeding to identify genotypes tolerant to water deficit. The aim of this study is to evaluate incorporation of seed weight (expressed as hundred kernel weight, HKW) as a covariate into genomic association and prediction studies for three biomass traits in a panel of elite inbred lines challenged by water withholding at seedling stage. Methods: 109 genotyped-by-sequencing (GBS) elite maize inbreds were phenotyped for HKW and planted in controlled conditions (16/8 day/night, 25 °C, 50% RH, 200 µMol/m2/s) in trays filled with soil. Plants in control (C) were watered every two days, while watering was stopped for 10 days in water withholding (WW). Fresh weight (FW), dry weight (DW), and dry matter content (DMC) were measured. Results: Adding HKW as a covariate increased the power of detection of associations in FW and DW by 44% and increased genomic prediction accuracy in C and decreased in WW. Conclusions: Seed weight was effectively incorporated into association studies for biomass traits in maize seedlings, whereas the incorporation into genomic predictions, particularly in water-stressed plants, was not worthwhile.
Traditional maize (Zea mays L.) breeding approaches use directly measured phenotypic performance to make decisions for the next generation of crosses. Indirect assessment of cultivar performance can be utilized using various methods such as genomic predictions and remote sensing. However, some secondary traits might expand the breeder’s ability to make informed decisions within a single season, facilitating an increase in breeding speed. We hypothesized that assessment of photosynthetic performance with chlorophyll a fluorescence (ChlF) might be efficient for in-season predictions of yield and grain moisture. The experiment was set with 16 maize hybrids over three consecutive years (2017–2019). ChlF was measured on dark-adapted leaves in the morning during anthesis. Partial least squares models were fitted and the efficiency of indirect selection was assessed. The results showed variability in the traits used in this study. Genetic correlations among all traits were mainly very weak and negative. Heritability estimates for all traits were moderately high to high. The model with 10 latent variables showed a higher predictive ability for grain yield (GY) than other models. The efficiency of the indirect selection for GY using biophysical parameters was lower than direct selection efficiency, while the indirect selection efficiency for grain moisture using biophysical parameters was relatively high. The results of this study highlight the significance and applicability of the ChlF transients in maize breeding programs.