US alfalfa variety trials show breeding progress for yield, broad adaptability of commercial varieties, and widespread genotype-by-trial variation

Public forage variety trials are the only objective source of information for alfalfa producers purchasing seed. There has been extensive research improving the efficiency of individual trials, but limited research guiding the best practices within a network of trials. A retrospective analysis of hundreds of alfalfa (Medicago sativa L.) trials in the US Midwest, Northeast, and Great Plains revealed that individual trials have limited external validity. Based on mean variety yield in a trial, only 36.2% of genetic variance was consistent across many trials. Genotype-by-environment effects explained 26.2% of the variance. Residual variance was confounded with genotype-by-trial variance and explained 37.6%. This resulted in standard errors of 4.1%, 3.3%, and 2.9% (variety deviation from trial mean yield) for varieties evaluated in two, three, and four trials, respectively. Practically, four or five trials are suggested as minimum threshold for a reliable yield estimate in alfalfa. Fortunately, the scale of genotype-by-environment variation is small relative to other crops, suggesting that superior alfalfa varieties are superior across a large geographic range. Last, there was breeding progress for biomass yield (+0.63% per year; standard error = 0.029%; p = 9.0 × 10⁻⁷⁸) in alfalfa between 1988 and 2017.