Prediction of seed longevity: a modified model and improved fitting process.

Abstract

This paper highlights the importance of estimates of seed longevity both for the management of germplasm collections and of commercial seed production and storage by introducing a model modified to allow for and estimate the initial proportion of the seed lot that was never viable. This model, based on the 'control mortality' probit model used in insecticide bioassays, includes an additional parameter to estimate the original viability of the seed lot (of cereal grains). Under this model, the slope and intercept of the response are independent of this original viability, and the time to reach 50% of the original viability is constant for given storage conditions. This modified model also allows for a loss of viability due to deterioration prior to testing and storage.