Abstract Soybean ( Glycine max , (L.) Merr.) productivity gains depend on accurate estimates of genetic trends, genotypic effects ( G ), and genotype-by-environment interactions (GEI). Here, we compiled a Value for Cultivation and Use (VCU) database of soybean varieties, evaluated from 2003 to 2018 in Austria (AT) and France (FR), to estimate genetic trends, G and GEI magnitudes, and variety-testing precision in grain yield, protein yield and protein content (%), using mixed-model methodology. We found that G and GEI variances were comparable in both countries, with a variety-testing precision, expressed as a relative detection threshold ( LSD for elite standards), of about 7.5% for grain yield and protein yield, and about 2% for protein content. Grain yield correlated moderately negatively with protein content at the genotypic level. Despite this, although genetic trends in protein content were non-significant and near-zero (slightly negative in FR), trends in grain yield and protein yield were significant and strongly positive (1.97–2.05% yr −1 , AT; 0.75–0.86% yr −1 , FR). Given the high heritability of protein content (> 0.9), there is scope for selection, which may come at the expense of grain yield. Both networks appeared well optimized in terms of the number of test locations, but gains in precision could be obtained from adding an extra test year. Combining data from both VCU networks enabled effective comparison between varieties not jointly tested. Such cross-country comparison could be further improved by strategic use of check varieties, genetic markers and environmental covariates to enhance genotypic and environmental connectivity between networks.
Ramakers et al. (Mon,) studied this question.