EVALUATION OF A CROP SIMULATION-MODEL THAT INCORPORATES GENE-ACTION

Crop simulation models are receiving increasing use in agricultural re search, However, applications to plant breeding have been limited, in part due to the restricted capabilities of models to represent genetic differences, A gene-based simulation model, GeneGro, has been develop ed that integrates action of seven genes into a common bean (Phaseolus vulgaris L.) model, The genes for physiological traits identified in the model include Ppd, Hr, Fin, Fd, Ssz-1, Ssz-2, and Ssz-3, Evaluatio n of GeneGro was performed by comparison of measured field data vs, si mulated data, and through a sensitivity analysis of the model, The exp erimental data set for model comparison was comprised of 14 field tria ls conducted in Canada, the USA, Mexico, and Colombia, representing 21 3 treatment combinations and including 39 cultivars, GeneCro explained 75% of variation in days to Bower, 68% in days to maturity, and 39% i n seed mass, but only 11% of variation in seed yield, Most of the vari ation in simulated seed yield was accounted for by the model when mean effects of site and cultivars were removed through regression analysi s. This suggests that the poor simulation of seed yield was partially due to errors in simulating differences among sites rather than from s imulating the response of genotypes to sites, Seed yield simulations w ere made for 96 genotypes that represented all expected phenotypic com binations of the seven genes included in the model, Simulations were c onducted using historical weather data from Fargo, ND; Madero, Mexico; and Palmira, Colombia, GeneGro predicted that photoperiod sensitivity , conferred by Ppd and Hr, would result in zero seed yields at Fargo, a slight seed yield increase at Madero, and no effect at Palmira. Test ing effects of maturity on seed yield, the genotypes fin and Fd result ed in lower seed yields due to shorter growth cycles, in agreement wit h field trials. The predicted effects of the three genes related to se ed size varied among the sites. These results support the use of genet ic information to represent cultivar differences in simulation models, They also, emphasize the need for a better understanding of the physi ological genetics of traits such as phenology, seed size, and growth h abit.