DEVELOPING GUIDELINES FOR REPLANTING GRAIN-SORGHUM .1. VALIDATION ANDSENSITIVITY ANALYSIS OF THE SORKAM SORGHUM GROWTH-MODEL

Sorghum [Sorghum bicolor (L.) Moench] producers often have difficulty determining when plant stands are low enough to merit replanting. Our objective was to use SORKAM, a sorghum growth model, to develop guidel ines for replanting grain sorghum. A necessary first step is validatio n of the model over an extensive range of management and environmental factors. Validation was accomplished using 19 field data sets represe nting 11 yr and six Kansas locations. Several nonparametric tests were used to compare observed and simulated yields, yield components, and phenological dates. In addition, sensitivities of yield and yield comp onents were determined in response to yearly climate, planting date, p lant population, and maturity class changes. Model sensitivities were compared with sensitivities calculated from field data. While phenolog ical predictions were adequate, SORKAM could capture only 27 to 79% of grain yield variability at the locations tested. Yield predictions fr om different plant populations within a planting date were particularl y inaccurate. The validation and sensitivity analyses showed that the poor yield predictions were the result of improper computation of till er number and faulty partitioning of biomass to caryopsis weight. Part itioning errors translocated enough assimilate from culm to grain to m ake yields essentially constant across populations within a planting d ate. To use SORKAM to generate replant guidelines, improvements must b e made in modeling the relationships among yield components and the so urce-sink relationship that determines caryopsis weight.