Assessing genotype adaptability and stability in perennial forage breeding trials using random regression models for longitudinal dry matter yield data.

Abstract

Genotype selection for dry matter yield (DMY) in perennial forage species is based on repeated measurements over time, referred to as longitudinal data. These datasets capture temporal trends and variability, which are critical for identifying genotypes with desirable performance across seasons. In this study, we have presented a random regression model (RRM) approach for selecting genotypes based on longitudinal DMY data generated from 10 breeding trials and three perennial species, alfalfa (Medicago sativa L.), guineagrass (Megathyrsus maximus), and brachiaria (Urochloa spp.). We also proposed the estimation of adaptability based on the area under the curve and stability based on the curve coefficient of variation. Our results showed that RRM always approximated the (co)variance structure into an autoregressive pattern. Furthermore, RRM can offer useful information about longitudinal data in forage breeding trials, where the breeder can select genotypes based on their seasonality by interpreting reaction norms. Therefore, we recommend using RRM for longitudinal traits in breeding trials for perennial species.