High-throughput phenotyping technologies, which can generate large volumes of data at low costs, may be used to indirectly predict yield. We explore this concept, using high-throughput phenotype information from Fourier transformed near-infrared reflectance spectroscopy (NIRS) of harvested kernels to predict parental grain yield in maize (Zea mays L.), and demonstrate a proof of concept for phenomic-based models in maize breeding. A dataset of 2,563 whole-kernel samples from a diversity panel of 346 hybrid testcrosses were scanned on a plot basis using NIRS. Scans consisted of 3,076 wavenumbers (bands) in the range of 4,000-10,000 cm −1. Corresponding grain yield for each sample was used to train phenomic prediction and selection models using three types of statistical learning: (a) partial least square regression (PLSR), (b) NIRS best linear unbiased predictor (NIRS BLUP), and (c) functional regression. Our results found that NIRS data were a useful tool to predict maize grain yield and showed promising results for evaluating genetically independent breeding populations. All model types were successful; functional regression followed by the PLSR model resulted in the best predictions. Pearson's correlations between predicted and observed grain yields exceeded .7 in many cases within random cross validation. Abbreviations: AF, aflatoxin; BLUE, best linear unbiased estimator; BLUP, best linear unbiased predictor; CV, cross validation; CV0, predicting one environment using data from all other environments; CV1, 20% of the hybrids are predicted by the remaining 80% of hybrids (five-fold), within each environment; CV2, predicting across environments, where hybrids are seen in some environments but predicted in others (mimics sparse testing); G × E, genotype by environment; G-BLUP, genomic best linear unbiased predictor; GEM, germplasm enhancement of maize lines; GWAS, genome-wide association study; LM, simple linear model; NIRS, near-infrared reflectance spectroscopy; NIRS BLUP, NIRS-based best linear unbiased predictor; PLSR, partial least squares regression; RMSEP, root mean square error of prediction; SERAT, southeast regional aflatoxin trial; UAS, unoccupied aerial systems; WS, water stress, unirrigated treatment; WW, well-watered, irrigated treatment. This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made.

Studies assessing phenotypes of plant populations traditionally place their primary focus on increasing measurement precision and improving accuracy. Phenotyping methods that use imaging, remote sensing, and spectroscopy, continue to increase throughput capacity, but information has been unavailable to assess the tradeoffs between increased throughput and any potential decreases in measurement accuracy. In this simulation study, we compare four levels of measurement accuracy across varying levels of throughput, and discuss how an increased error rate can be compensated for via increased throughput, if experimental resources are allocated appropriately. Under the tested scenarios of increased throughput, the correct values of genotypes were best estimated by increasing the number of environments.Genetic mapping studies should increase population size as well to see improvements over more accurate measurement methods. This simplistic simulation mimics many empirical findings and will be of interest to any researcher interested in assessing how high-throughput phenotyping methods affect decision-making in crop research programs.Abbreviations: BLUE, best linear unbiased estimator; DH, doubled haploid; G × E, genotype × environment; HTP, high-throughput phenotyping; HTP-2×, high-throughput method with the error variance twice the size of the genetic variance; HTP-5×, high-throughput method with the error variance five times the size of the genetic variance; HTP-10×, high-throughput method with the error variance 10 times the size of the genetic variance; NIRS, near-infrared reflectance spectroscopy; QTL, quantitative trait locus; SNP, single nucleotide polymorphism; UAS, unoccupied aerial system.This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.