Variability in flower development of Easter lily ( Lilium longiflorum Thunb.): model and decision-support system

A model to predict the distribution of harvest dates in an Easter lily crop was validated using data from several locations. Plants were individually harvested (i.e. removed from the greenhouse for shipping) when flowers on a plant reached a minimum flower bud length. A computer decision-support system called LilyDate was developed to allow the model to be used to optimize greenhouse temperature settings to ensure that the majority of a crop is ready to harvest by a target date. To implement LilyDate , the user measures the length of the largest flower bud per plant on a sample of plants in the greenhouse, and enters the frequency of plants at each flower bud length (to the nearest cm) into the program. Based on the expected temperature and sample growth data, the software predicts a cumulative and daily distribution of when plants will be in flower. Model predictions were not biased, and the model predicted the number of days until 50 or 95% of plants were in flower with a precision of ±2 and ±3 days, respectively. The general approach to the computer system could be adapted to predict the harvest distribution for other crop species that require a consistent quantity of thermal time to harvest, and that are grown for a target harvest date.