A simple model for crop management applications for cotton (Gossypium hirsutum L.)

Abstract SIRATAC is a computer-based tactical cotton pest management system in commercial use in Australia. Initially, SIRATAC did not include a crop model, but during the period 1976 to 1983 a simple dynamic simulator of cotton fruiting evolved through the following stages: 1. (1) Fruit was counted regularly and compared to a non-dynamic trajectory of development set by the manager in order to determine whether the crop was likely to achieve his expectations. 2. (2) A function expressing fruit shedding in terms of fruit load was introduced in order to predict the number of counted fruit likely to contribute to yield. 3. (3) Using this function with the heat sum requirement for fruit development and data collected by SIRATAC, a function was developed to predict production of new fruiting sites in terms of fruit load and the cumulative number of fruiting sites. 4. (4) The concept of carrying capacity (the fruit load that reduces fruit survival and fruiting site production to zero) was developed. The concept implicitly incorporates the carbon economy of the crop; the ratio of carrying capacity to fruit load is a surrogate for the ratio of carbon supply to carbon demand. The result of these developments was a simple simulation model consisting of three processes (fruit site production, fruit shedding, and fruit heat sum requirements) and four state variables, of which three (squares, bolls, and open bolls) are readily counted and the other (fruiting sites) is readily derived during simulation. Age profiles of counted and predicted fruit are also generated during the simulation. An iterative procedure was developed to calibrate the shedding and site production functions from the data collected during the commercial use of the model. The current version of the model was calibrated with 1979–1980 and 1980–1981 data and validated with data from 1981–1982 and 1982–1983. The model reflects the dynamics of cotton fruiting, responding realistically to loss of fruit by pest attack at different stages in accord with documented results. The model has been developed for within-season tactical predictions when data (counts and temperatures) are available up to the time at which a projection is made, in contrast to retrospectively mimicking a past season with all available meteorological data and without use of crop data. Calibration is specific for location and production system, but the model is readily calibrated to any locality when a season's counts of squares and bolls are available. The model assumes average nitrogen and water applications and that the soil and the range of weather patterns experienced during calibration will prevail during application of the model, which is reasonable for Australian cotton production.