Association of digital photo parameters and NDVI with winter wheat grain yield in variable environments

The normalized difference vegetation index (NDVI) is gaining popularity as a complementary selection tool, even though it requires an instrument not readily available in the developing world. We evaluated several parameters (originating from the analysis of digital photos using BreedPix software) as potential selection criteria in 23 winter wheat yield trials grown over 4 years at 2 sites. NDVI and digital photos were taken at key development stages from stem elongation to maturity. The correlations between digital photo parameters a and u and grain yield, as well as correlations between NDVI and grain yield within individual trials, varied depending on crop stage, moisture availability, and germplasm composition. NDVI, photo-a, and photo-u parameters had equal power in distinguishing high- and low-yielding genotypes in the trials and were significantly associated with yield in approximately 50% of all observations. The association of vegetative indices with grain yield can be improved by evaluating germplasm with a similar maturity range. An important challenge is in utilizing these tools in unreplicated small plots, including head rows where selection efficiency is low.