Temporal phenomic predictions from unoccupied aerial systems can outperform genomic predictions.
G3 (Bethesda)
; 13(1)2023 01 12.
Article
em En
| MEDLINE
| ID: mdl-36445027
ABSTRACT
A major challenge of genetic improvement and selection is to accurately predict individuals with the highest fitness in a population without direct measurement. Over the last decade, genomic predictions (GP) based on genome-wide markers have become reliable and routine. Now phenotyping technologies, including unoccupied aerial systems (UAS also known as drones), can characterize individuals with a data depth comparable to genomics when used throughout growth. This study, for the first time, demonstrated that the prediction power of temporal UAS phenomic data can achieve or exceed that of genomic data. UAS data containing red-green-blue (RGB) bands over 15 growth time points and multispectral (RGB, red-edge and near infrared) bands over 12 time points were compared across 280 unique maize hybrids. Through cross-validation of untested genotypes in tested environments (CV2), temporal phenomic prediction (TPP), outperformed GP (0.80 vs 0.71); TPP and GP performed similarly in 3 other cross-validation scenarios. Genome-wide association mapping using area under temporal curves of vegetation indices (VIs) revealed 24.5% of a total of 241 discovered loci (59 loci) had associations with multiple VIs, explaining up to 51% of grain yield variation, less than GP and TPP predicted. This suggests TPP, like GP, integrates small effect loci well improving plant fitness predictions. More importantly, TPP appeared to work successfully on unrelated individuals unlike GP.
Palavras-chave
Texto completo:
1
Coleções:
01-internacional
Base de dados:
MEDLINE
Assunto principal:
Estudo de Associação Genômica Ampla
/
Fenômica
Tipo de estudo:
Prognostic_studies
/
Risk_factors_studies
Limite:
Humans
Idioma:
En
Revista:
G3 (Bethesda)
Ano de publicação:
2023
Tipo de documento:
Article