A stochastic process modelling of maize phyllochron enables to characterize environmental and genetic effects.

The times between appearance of successive leaves or phyllochron characterize the vegetative development of annual plants. Hypothesis testing models, which enables to compare phyllochron between genetic groups or conditions, are usually based on regression of thermal time on the number of leaves, most of the time assuming a constant leaf appearance rate. However these models are both statistically biased and inappropriate in terms of modelling. We propose a stochastic process model in which the emergence of new leaves is considered as successive time-to-events, which provides a flexible and more accurate modelling as well as unbiased testing procedures. The model was applied on an original maize dataset collected in fields for three years on plants originating from two divergent selection experiments for flowering time conducted in two maize inbred lines. We showed that the main differences in phyllochron were not observed between selection populations (Early or Late), but rather between ancestral lines, years of experimentation, and leaf ranks. Our results highlight a strong departure from the assumption of a constant leaf appearance rate in one year that could be related to climate variations, even if the impact of each climatic variables individually was not clearly elucidated.