Abstract :
The research objectives were to improve the probit model for predicting longevity of orthodox seeds by taking into account the seed composition, and to conduct an error analysis to determine the uncertainties involved in the probit model for three seed species (chickpea, cowpea, and soya bean). Multiple linear regression method was employed to determine the dependency of the viability constants of the probit model on seed composition. Probit equations, already developed for six seed species (barley, chickpea, cowpea, groundnut, sorghum, and soya bean), were used to generate a data set in the ranges of 5–25% moisture content and −20 to 70 °C temperature. The improved model obtained, called a generalised longevity model, can be used to predict seed viability as a function of initial seed viability, time period, and seed temperature, moisture content, and composition (carbohydrate, lipid, and protein fractions). The average error for predicting seed viability by using the proposed model (5·8%) was much lower than the average uncertainty of the probit equations (33·9%). The generalised longevity model could be used for seed species with known composition and unknown viability constants. Further research is recommended to validate the proposed model by using experimental data of different seed species.
