A novel GIS-based tool to reveal spatial trends in reaction norm: upland rice case study

Abstract

The upland rice crop system located within Brazilian savannas and Amazon Rainforest is the largest rainfed rice growing area in Latin America. To develop and release higher yield and adapted cultivars for this large region, the upland rice breeders need to conduct multiple-location trials aiming to model the genotype × location (G × L) and evaluate the germplasm yield adaptability. Here we hypothesize that regional patterns of G × L across this extensive region can be modeled by integrating factorial regression models with a geographic information system (GIS). Two sets of advanced yield trials from different germplasm pool were used in this study. From GIS tools, we collect and process geographic covariates and produce thematic maps of yield adaptability. One advantage of the methodology is that adaptability can be dissected into genotypic-sensibility coefficients related to the reaction norm for the geographic gradient. Then, breeders can discriminate different types of adaptability over a region, such as responsiveness for elevation, longitudinal or latitudinal adaptation, identifying possible ideotypes to solve current adaptation gaps for target regions. We observed that about of 53–59% of the G × L effects are due to predictable geographic-related factors. However, the upland rice germplasm is better adapted to higher elevations (> 700 m), which may indicate limitations in cultivar development because these regions do not represent the current upland rice growing region. We suggest to exploit geographic-related factors by increasing breeding efforts for northern and western Brazil environments located at lower elevations (< 300 m) and Equador’s near latitudes (2° S–2° N).