Adapting the Cropgro Model to Predict Growth and Yield of Pigeonpea (Cajanus cajan).

The objective of this study was to adapt the CROPGRO model for simulating pigeonpea (Cajanus cajan L. Millsp.) growth and development.  The model is process-based and predicts growth and yield of a number of other grain legumes.  Parameters for plant growth and development are stored in external read-in files to facilitate adaptation of the model for new species, without changing the generic source code.  Parameter values for pigeonpea were determined via literature review and parameter estimation based on data sets from Florida and Madhya Pradesh, India.  A version of the Metropolis-Hastings algorithm was used for parameter estimation by optimization against growth analyses data.  Relative root mean square error (RRMSE), Wilmott agreement index (D-stat), and Nash-Sutcliffe efficiency (C_eff) were used as model evaluation criteria.  Model simulation of dynamics of biomass and plant components over time showed good performance with mostly high (> 0.90) D-stat and C_eff values and RRMSE values generally less than 0.20.  Leaf area index and specific leaf area showed good correspondence between observed and predicted values.  Likewise, CROPGRO predicted pigeonpea phenology well.  Nitrogen concentration dynamics were less well captured by model simulations (D-stat < 0.80; C_eff < 0.40).  This adaptation showed that CROPGRO can successfully predict pigeonpea growth and development.  Further evaluation with independent datasets is needed to confirm these findings.