Improvement on The Ellis and Roberts Viability Model

With data sets of germination percent and storage time of seed lot of wheat and sorghum stored at three different storage temperature(t, °C) with three different water content (m, %) of seeds, together with data set of buckwheat and lettuce reported in literatures, the possibility that seed survival curve were transformed into line by survival proportion and the relationship that logarithm of average viability period (logp50) and standard deviation of seed death distribution in time (δ)with t, m and interaction between t and m were analysed. Result indicated that survival proportion transformed seed survival curve to line were much easier than the probability adopted by Ellis and Roberts, and the most important factor affecting logp50 and δ of seed lot was interaction between t and m. Thus, Ellis and Roberts viability model were suggested to be improved as Ki=Vi-p/10K-CWT (t×m) to predict longevity of seed lot with initial germination percent unknown, a new model of Gi/G0=A-P/10K-CWT(t×m) was constructed to predict longevity of seed lot with initial germination percent already known.