Current methodology to quantify temperature characteristics in germination of seeds is predominantly based on analysis of the time to reach a given germination fraction, that is, the quantiles in the distribution of the germination time of a seed. In practice interpolation between observed germination fractions at given monitoring times is used to obtain the time to reach a given germination fraction. As a consequence the obtained value will be highly dependent on the actual monitoring scheme used in the experiment. In this paper a link between currently used quantile models for the germination time and a specific type of accelerated failure time models is provided. As a consequence the observed number of germinated seeds at given monitoring times may be analysed directly by a grouped time-to-event model from which characteristics of the temperature profile may be identified and estimated. Simulations indicate that the performance of the proposed methodology is satisfactory irrespective of the actual monitoring scheme. Finally, the model is applied to a quinoa germination experiment studying the impact of soil salinity.
Journal of Agricultural, Biological, and Environmental Statistics, 2013, Vol 18, Issue 1, p. 87-101