University of Florida, Citrus Research and Education Center, 700 Experiment Station Road, Lake Alfred, FL 33850, USA

Background and objectives
Post bloom fruit drop (PFD) of citrus was first reported in Belize in 1979 and has since become a problem in virtually all humid growing areas in the Americas [1]. Conidia of Colletotrichum acutatum infect citrus petals, produce reddish-brown lesions and the fungus subsequently sporulates profusely on the petal surface. After infection, the fruitlet drops but the calyx and floral disc remain attached, forming structures commonly called buttons. The fungus persists between bloom periods as appressoria formed on leaves, twigs and buttons. In spring, petal exudates stimulate germination of appressoria and the formation of a few conidia to complete the life cycle. The disease is controlled by applications of benomyl and other fungicides during bloom.

The objectives of these studies were to evaluate the environmental effects on disease incidence, develop a predictive model for PFD, and validate the usefulness of the model in disease management.

Materials and methods
Disease incidence and environmental conditions were monitored biweekly for two seasons. Stepwise multiple regression analysis was used to relate environmental factors to disease incidence from which a predictive model was developed. The model was evaluated for 3 years in three to five locations to determine its usefulness in predicting the need for fungicide applications, as well as proper timing [2].

Results and conclusions
The most significant factors affecting disease incidence were: (1) inoculum represented by the number of diseased flowers per tree, (2) rainfall and (3) leaf wetness. Temperature and relative humidity were not significant factors affecting disease incidence. The following formula was used for predictions:
where y=predicted percentage of diseased flowers 3-4 days later; TD=the total number of affected flowers on 20 trees; R=rainfall for the last 5 days in mm. If TD<75, then TD was considered to be zero. Fungicide applications were considered necessary if the model predicted >20% incidence because groves with <20% disease never experienced yield loss. Utilization of the model in 1993 in three groves, triggered one to three fungicide applications and fruit increases of 53-261%; in 1994, from none to two sprays were indicated in five groves with fruit increases of 25-65%; in 1995, from none to two sprays were indicated in five groves with fruit increases of 52-523%. Where sprays were not indicated, disease levels remained low and no losses were incurred. The model has recently been adjusted to include leaf wetness information now available to many growers. The model has been very useful for timing fungicide applications in seasonally dry, subtropical Florida. In humid tropical areas, where inoculum is commonly present at high levels and rainfall is frequent during bloom, the model provides little benefit.

1. Timmer LW, Agostini JP, Zitko SE, Zulfiqar M, 1994. Plant Disease 78, 329-34.
2. Timmer LW, Zitko SE, 1996. Plant Disease 80, 166-69.