NDVI-based ideotypes as a cost-effective tool to support wheat yield stability selection under heterogeneous environments

Yield stability is a trait of increasing importance for food security given climate change, particularly in environments with diverse growing conditions and extreme abiotic stressors. Yet, stability is rarely investigated in early stages of a breeding pipeline due, in part, to the insufficient identification of secondary traits needed to inform selection decisions. The objective of this study was to develop an NDVI-based phenotype capable of identifying yield-adaptive traits for common wheat ( T. aestivum ) grown under heterogeneous environments, including conditions of nitrogen (N) deficiency and terminal drought. In 26 trials conducted during three seasons across the primary wheat-growing regions of California (United States), we measured yield for 25 common wheat genotypes in highly variable production settings. Multitemporal UAV-based NDVI measurements were recorded from a single location-year-environment combination (n = 3 of 26, hereafter: representative environment, i.e. ‘RepEnv’) and combined to develop a 4-part, seasonal reflectance phenotype per genotype. Groupwise ideotypes were developed from these phenotypes to identify yield-adaptive traits under average yield, N deficiency, and terminal drought stress conditions. Phenotypes and yields measured in the RepEnv were used to predict statewide yield stability outcomes and identify adaptive traits at the genotype level. Although yield and yield stability were negatively correlated, ideotypes for high yield and high yield stability were in partial agreement, with early-season vigor contributing to both high yield and yield stability compared to low-yielding and low-stability ideotypes (p = 0.02 and 0.20, respectively). Whereas, late-season growth patterns differentiated high versus low stability and yield adaptation to N deficiency versus terminal drought stress. At the genotype level, cross-validated predictions of stability made using the NDVI-based phenotypes were linearly correlated with observed values both alone (p = 0.07) and when combined with yield measurements made in the RepEnv (p = 0.01). Further, the quantitative relationship between observed and phenotype-predicted stability values enabled identification of genotypes with adaptive traits, including positive deviations. The UAV-based 4-part NDVI phenotype developed is a useful, low-cost tool to support selection of stable and stress-resilient wheat genotypes in heterogeneous environmental conditions, promoting efficient breeding and contributing to long-term food security. • NDVI-based ideotypes of yield and yield stability quantitatively illustrate adaptive traits. • Early vigor improved both yield and yield stability under diverse environmental conditions. • Late-season growth patterns differentiated adaptation to N deficiency vs terminal drought stress. • Combining NDVI phenotype with yield data supported identification of adaptive traits.