Résumé
Effects of photoperiod (PP) on panicle initiation (PI) of West African sorghums are essential for the crop's agro-ecological adaptation, but not well understood. Conventional models based on fixed PP thresholds (qualitative response) or additive signal accumulation (qualitative response) are able to predict flowering for only a limited range of conditions. Recently, an alternative concept was proposed based on thresholds that vary with plant age. Using this concept, this study develops a generic model of PP response of sorghum called "Impatience" because it implements decreasing day length requirements during prolonged wait states, or appetence, during the photoperiod sensitive phase (PSP).
The model was applied to experimental field data obtained from sowing date experiments in Mali, in order to validate it and evaluate the range of genotypic response that can be explained with it. It predicted accurately the observations that (1) PI does not occur at any genotype specific day length, but instead, on increasingly long days as the PSP is extended; (2) PI occurs predominantly when day length decreases, or after summer solstice; (3) the duration of PSP increases linearly (but not always proportionally) when crops are sown earlier in the year; (4), a genotype specific sowing date exists in winter (cool season) or spring (hot dry season) after which PSP suddenly increases by up to 160 days ("break point"); (5), the largest variance of PSP occurs near the break point. Most genotypic variations of this complex pattern could be simulated by fitting only one of the four model parameters while using default values for the three other parameters. However, the model tended to underestimate the duration of PSP when it fell into the cool season, which might be explained by specific thermal effects or a specific inhibitory effect of increasing day length on PI in some genotypes.
The ability of the model to discriminate among genotypes was used to devise a simple, model assisted phenotyping methodology. Genotypic parameters were determined by fitting two model parameters to measured dates of flag leaf ligule appearance for three sowing dates. Thus parameterized, the model predicted accurately the phenological patterns of contrasting materials observed in different experiments. The authors discuss the physiological probability of the model's underlying concepts, the need for further studies to develop them into a comprehensive theory, and possible applications in crop simulation and varietal selection. (c) 2007 Published by Elsevier B.V.