Assessing Impact of Trait Modification Using Crop Models.

Empirical, field tests of alterations in plant performance resulting from trait modification are ultimately limited to the environments in which the germplasm is tested. Total seasonal rainfall, for example, has proved not to be a situable yield predictor across seasons. An alternative to assess the putative benefit of a trait across locations and a number of seasons is to use mechanistically based crop models that track soil water availability and crop response through the growing season. However, there are also some critical issues related to the use of some of these mechanistic models. Many models use meteorological functions, including the Penman equation, to estimate crop water loss. The Penman equation may not be readily transferable to most cropping environments, and consequently, in these models the equation is diminished to an empirical expression usually including a 'crop coeffcient'. Further, complex models requiring a number of descriptive coefficients are difficult to implement and the coefficients are commonly not extrapolative across a wide range of environments. An alternative is to use comparatively simple models that have many fewer coefficients and do not rely on a meteorological estimates of water loss. We have undertaken such a model assessment for several candidate traits for soybean yield increase across environments. Simulations were done for traits using 50 years of weather data in 30 km x 30 km grids across the cropping regions of the U.S. Results from these simulations indicated the probability of yield gain for each location, and the extent of yield increase or decrease that might be anticipated.