Statistical studies using AIC method to decide the question of "break" or "straight" in arrhenius plots of water proton NMR relaxation times in chilling-sensitive vigna and insensitive pisum seedlings

The question of "break" or "straight" in Arrhenius plots for the temperature dependency of NMR relaxation times (T₁) of water protons in etiolated intact seedlings for chilling-sensitive two Vigna species and chilling-insensitive Pisum was statistically studied using the Akaike's Information Criterion (AIC), a versatile procedure for statistical model identification. Among seven models, the most appropriate was based on the following assumptions: individual k (preparation numbers) two-half lines connecting at the break point (TCB) and imposing no restrictions for the gradient and the break point (Model 7). The worst two models were one straight line or one TCB (Model 1 or 4). Thus, the data obtained from replicated preparations should not be treated as a whole but as individual in each sample case. Break points determined with Model 7 ranged around 11-5°C for V. radiata and 17-10°C for V. mungo. The implication of the occurrence of "break" for Pisum clearly differed from Vigna judging by the model fitness based on the AIC values. The question of "break" or "straight" in Arrhenius plots is therefore fairly dependent on the validity of the model selection in the statistical analysis, and the AIC method is a useful procedure for the resolution of the problem of the "use" or "misuse" of Arrhenius plots in plant physiology.