Enviromic prediction is useful to define the limits of climate adaptation: A case study of common beans in Brazil

Abstract

Future environmental shifts foster plant research aiming to develop climate-smart cultivars but the past and current impacts on the environment are also a key to unraveling a major part of the phenotypic adaptation of crops. These studies may determine the most relevant environmental components of yield stability and adaptability within a breeding framework. Here, as a proof-concept study, we quantified the impacts of climate drivers in adapting common bean across Brazilian regions and seasons. We developed an enviromic prediction approach based on Generalized Additive Models (GAM), large-scale environmental covariate data (EC), and grain yield (GY) of 18 years of a common bean breeding program. Then, we predicted the optimum limits for ECs for each production scenario. We verified the ability of GAM-based models to explain the climate driver GY variation and performed accurate predictions for diverse production scenarios (four regions, three seasons, and two grain types). Our results indicates that air temperature (maximum and minimum), accumulated solar radiation, and rainfalls are mostly associated as the main drivers of GY variation in most regions. We also observed a huge variability of the climate drivers impact for the same germplasm cultivated across different seasons for each region. Furthermore, this climate influence in common beans adaptation is more evident during the vegetative for some seasons, while more impressive for reproductive stages for other seasons. Consequently, it demands higher efforts from breeding programs in developing region- or season-specific ideotype cultivars. Enviromics prediction with GAM was useful to identify the effect of climate on critical crop stages, which indirectly might help breeders in developing climate-smart varieties. We envisage its use with research field trial data (e.g., advanced yield testing) and historical farm field yield aimed at understanding breeding gaps in developing adapted cultivars for growing scenarios. HighlightsO_LIWe developed an enviromic prediction approach based on Generalized Additive Models (GAM) for a large-scale environmental covariate data and grain yield C_LIO_LIWe verified the ability of GAM-based models to explain the climate driver grain yield variation and performed accurate predictions for diverse production scenarios (four regions, three seasons, and two grain types) C_LIO_LIClimatic limitations for cropping commun beans were identified across seasons and regions. C_LIO_LI"Optimum" values for climate variables in different common bean regions productions were obtained . C_LI