A Statistical Perspective on Threshold Type Germination Models

Abstract

A need for empirical germination models that would give reliable predictions under field conditions led us to look critically at the statistical assumptions made in the concepts of thermal and hydrothermal time. Doubts were raised about the unquestioning description of rate temperature relationships by straight lines and the subsequent extrapolation of base temperatures from them. An alternative method where base temperatures are estimated from final percentage germination is suggested. This removes the linearity constraint from the rate temperature relationship and should have better predictive properties at low temperatures. An advantage of the method is that it predicts both the timing and percentage of seeds which will germinate in a given time interval. Thus it provides a unifying hypothesis which may aid interpretation of results from alternating temperature studies. Generalization of the method to include water potential is described and the methods are illustrated with several data sets.