Unravelling complex traits in wheat: approaches for analyzing genotype by environment interactions in a multi-environment study of falling numbers

Authors: SJOBERG, STEPHANIE - Washington State University; CARTER, AARON - Washington State University; Steber, Camille; Garland-Campbell, Kimberly

Submitted to: Crop Science
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 2/17/2020
Publication Date: 2/17/2020
Citation: Sjoberg, S.M., Carter, A.H., Steber, C.M., Garland Campbell, K.A. 2020. Unravelling complex traits in wheat: approaches for analyzing genotype by environment interactions in a multi-environment study of falling numbers. Crop Science. 60,6:3013-3026. https://doi.org/10.1002/csc2.20133.
DOI: https://doi.org/10.1002/csc2.20133

Interpretive Summary: The wheat industry uses the Hagberg-Perten falling number (FN) test to measure starch degradation caused by alpha-amylase enzyme activity in flour (Perten, 1964; Yu et al., 2015). Alpha-amylase is produced during seed germination to mobilize stored reserves to fuel seedling growth. However, flour with high alpha-amylase resulting in low falling numbers tends to produce poor quality baked goods, such as cakes that fall and sticky noodles. Thus, farmers receive steeply discounted prices for low FN grain. To help farmers choose varieties less prone to low FN, over 12,000 FN data points have been collected by researchers from the USDA-ARS and Washington State University, and made publicly available. While farmers would like to have this data converted into a relative ranking of varieties for risk of low FN, statistical analysis is complicated by the fact that low FN is caused both by preharvest sprouting and LMA, and by the fact that different varieties are grown in different locations and years by Cereal Variety Testing. This paper explores statistical methods to objectively rank varieties for their tendency towards low FN over a large dataset.

Technical Abstract: Multi-environment trials provide useful information about highly variable, complex plant traits like yield and quality, but are difficult to analyze due to their unbalanced nature, with genotypes and locations varying from year to year. Multiple approaches for characterizing relative genetic tendencies were explored across a three year multi-environment wheat variety trial dataset, examining falling numbers (FN) test results. The FN test measures the decrease in flour gelling capacity resulting from starch digestion by the enzyme alpha-amylase. Low FN/high alpha-amylase grain is discounted in the wheat industry because it is associated with poor end-use quality. The utility of methods employing mega-environments to assess relative variety performance was examined because low FN can be caused by susceptibility either to preharvest sprouting when it rains before harvest or to late maturity alpha-amylase induction by temperature fluctuations during grain maturation. The most effective approach for selecting varieties with stably high FN was a combination of joint regression, such as Finlay-Wilkinson and Eberhart and Russell, with biplot methods such as the additive main effects and multiplicative interaction model (AMMI) and the genotype main effects and genotype x environment interaction model (GGE).