Uncertainty in Plant Traits: Integrating Crop Water Deficit Field Experiments with Crop Modeling.

While trait modeling is receiving increasing emphasis in quantitative plant modeling, difficulties exist in 1) measuring the trait, and 2) integrating the measurements of the trait into quantitative plant models.  Measuring traits, particularly under carefully controlled/identified field conditions, inheritantly has error and several sources of variation.  Uncertainty exists about how the trait varies within a genotype across environments and among genotypes.  Integrating plant traits into our plant process models normally is via plant parameter inputs, usually as a fixed discrete value.  Often  sufficient data are not available  for determining the parameter value.  Ideally, the uncertainty of the parameter value could be captured by distributions of the parameter for different genotypes in different environments, and then use Monte Carlo simulation to sample from the distributions in our simulation models.  This paper examines phenological data (and possibly other data such as final plant height) for 24 winter wheat genotypes to create distributions for environments differing in water availability.  To do this, we use data from a 3-yr field experiment having five levels of water deficits precisely controlled by drip irrigation in northeast Colorado.  The phenological data include measurements of the date of jointing, flag leaf blade complete, heading, anthesis, and physiological maturity.  The irrigation scheduling includes full irrigation and dryland treatments, with irrigation in some dryland treatments at specific developmental stages (e.g., jointing and anthesis).