Quantitative comparisons of three modeling approaches for characterizing drought response of a highly variable, widely grown crop species
Abstract
Quantifying the drought tolerance of crop species and genotypes is essential in order to predict how water stress may impact agricultural productivity. As climate models predict an increase in both frequency and severity of drought corresponding plant hydraulic and biochemical models are needed to accurately predict crop drought tolerance. Drought can result in cavitation of xylem conduits and related loss of plant hydraulic conductivity. This study tested the hypothesis that a model incorporating a plants vulnerability to cavitation would best assess drought tolerance in Brassica rapa. Four Brassica genotypes were subjected to drought conditions at a field site in Laramie, WY. Concurrent leaf gas exchange, volumetric soil moisture content and xylem pressure measurements were made during the drought period. Three models were used to access genotype specific drought tolerance. All 3 models rely on the Farquhar biochemical/biophysical model of leaf level photosynthesis, which is integrated into the Terrestrial Regional Ecosystem Exchange Simulator (TREES). The models differ in how TREES applies the environmental driving data and plant physiological mechanisms; specifically how water availability at the site of photosynthesis is derived. Model 1 established leaf water availability from a modeled soil moisture content; Model 2 input soil moisture measurements directly to establish leaf water availability; Model 3 incorporated the Sperry soil-plant transport model, which calculates flows and pressure along the soil-plant water transport pathway to establish leaf water availability. This third model incorporated measured xylem pressures thus constraining leaf water availability via genotype specific vulnerability curves. A multi-model intercomparison was made using a Bayesian approach, which assessed the interaction between uncertainty in model results and data. The three models were further evaluated by assessing model accuracy and complexity via deviance information criteria (DIC). Results suggest that model 1 was unable to model soil moisture accurately and thus did not effectively characterize drought tolerance. Models 2 and 3 were both effective at characterizing drought tolerance; model 3 preformed best in genotypes with the highest vulnerability to cavitation. By identifying through both Bayesian and DIC analyses models that best characterize drought tolerance future investigations into the interaction between crop productivity and water use can be informed by hypothesis testing using models prior to experimentation.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2013
- Bibcode:
- 2013AGUFM.B11E0406P
- Keywords:
-
- 0466 BIOGEOSCIENCES Modeling;
- 0402 BIOGEOSCIENCES Agricultural systems;
- 1847 HYDROLOGY Modeling;
- 1852 HYDROLOGY Plant uptake