Development of statistical models for analysis of seed germination behaviour of rice ( <scp> <i>Oryza sativa</i> </scp> L.) under differential temperature and water potential

Abstract Seed germination testing is an important aspect to assess the quality of seed. Response to germination is a complex phenomenon governed by genetic and physio‐biochemical factors and environmental factors such as temperature, water potential and so forth. The germination time course is modelled with temperature as a covariate when imbibed under saturated conditions by using the thermal time property of seeds. However, the thermal time model does not account for a threshold effect of water potential in the situation of incomplete germination. Therefore, properties of the three‐parameter log‐logistic distribution have been explored theoretically to satisfy hydro time properties of seed germination when modelling this process. The parameters could differentiate the effects of biological factors on seed germination, which will help the investigation of rice seed viability in various climatic conditions. A novel methodology has been developed for efficient estimation of parameters that accounts for correlation and heteroscedastic error variance. The performance of the estimation methods, along with the maximum likelihood estimation method, has been studied. Finally, the impact of plant growth and seed maturity conditions in the mother plant upon seed germination has been investigated theoretically. Unlike the estimation of only cardinal temperature which does not account for a threshold effect of water potential, favourable temperatures for rice varieties of various maturity groups germinated under different water stress conditions have been recommended in the current research.