Prediction and mapping the productivity of eucalyptus clones with environmental covariates

Abstract

The quantitative nature of wood production poses a challenge for breeders. The complex interaction of genotypes with environments (G×E) makes cultivars recommendation difficult. Our objective was to model the G×E interaction using environmental covariates and map the adaptability of commercial Eucalyptus clones based on a geographic information system (GIS) across important plantation regions in Brazil. To achieve this, a productivity dataset with 13,483 stands of six commercial clones was utilized. The effects of geography, soil and climate covariates on clone yield were modeled using partial least squares regression, with data from WorldClim and SoilGrids databases. Using the models for each clone, yield maps were generated at a spatial resolution of approximately 5 km². Then, cultivar recommendation was made through a pixel-by-pixel comparison of predicted yield values among the clones. The covariates that most affected the performance of the clones were annual rainfall, rainfall of the driest month, rainfall of the driest quarter, maximum temperature of the hottest month and average temperature of the wettest quarter. Thus, G×E modeling based on environmental covariates combined with GIS enables a large increase in the resolution of cultivar recommendations by mapping the adaptability of genotypes in each site.