The development of a new crop growth model SwitchFor for yield mapping of switchgrass

Abstract

Switchgrass is a promising energy crop has the potential to mitigate global warming and energy security, improve local ecology and generate profit. Its quantitative traits, such as biomass productivity and environmental adaptability, are determined by genotype-by-environment interaction (GEI) or response of genotypes grown across different target environments. To simulate the yield of switchgrass outside its original habitat, a genotype-specific growth model, SwitchFor that captures GEI was developed by parameterising the MiscanFor model. Input parameters were used to describe genotype-specific characteristics under different soil and climate conditions, which enables the model to predict the yield in a wide range of environmental and climate conditions. The model was validated using global field trail data and applied to estimate the switchgrass yield potentials on the marginal land of the Loess Plateau in China. The results suggest that upland and lowland switchgrass have significant differences in the spatial distribution of the adaptation zone and site-specific biomass yield. The area of the adaption zone of upland switchgrass was 4.5 times of the lowland ecotype's. The yield difference between upland and lowland ecotypes ranges from 0 to 34 Mg ha−1. The weighted average yield of the lowland ecotype (20 Mg ha−1) is significantly higher than the upland type (5 Mg ha−1). The optimal yield map, generated by comparing the yield of upland and lowland ecotypes based on 1 km2 grid locations, illustrates that the total yield potential of the optimal switchgrass is 61.6–106.4 Tg on the marginal land of the Loess Plateau, which is approximately twice that of the individual ecotypes. Compared with the existing models, the accuracy of the yield prediction of switchgrass is significantly improved by using the SwitchFor model. This spatially explicit and cultivar-specific model provides valuable information on land management and crop breeding and a robust and extendable framework for yield mapping of other cultivars.