Istituto di Entomologia e Patologia vegetate, Universita Cattolica S. Cuore, Via E. Parmense 84, 291 00 Piacenza, Italy

Background and objectives
Downy mildew of onion, caused by Peronospora destructor (Burk.) Casp. (syn. P. schleideni Unger), has a wide geographical distribution and causes severe outbreaks in Italy, especially in the northern area. P. destructor is a polycyclic pathogen: many infection cycles can follow one another during an onion-growing season. When weather conditions are favourable, the fungus can complete its cycle in a short time and the disease can cause severe yield losses. Fungicidal applications are always necessary to guarantee crop health and they are usually applied prophylactically: they start before disease onset and go on till 4-5 applications according to a regular schedule. A model able to forecast primary infections would be a useful tool in improving disease management and in reducing early fungicide applications.

Materials and methods
Systems analysis was applied to the onion-downy mildew pathosystem in order to build up a computerized simulation model for primary infections. Data from the literature were used to define mathematical relationships between weather and phases of the infection cycle, with particular emphasis on sporulation on inoculum sources, spore survival, infection establishment, and incubation. In order to validate the model, simulations were compared to actual data collected in the field at several locations during a 5-year period (1992-96). Experiments were carried out in 1997 to study the influence of primary inoculum and the growth stage of onion on infection establishment. The former was monitored by means of spore traps or by looking at plants grown from infected bulbs in order to detect presence of the disease symptoms; the latter was considered by means of various seeding times.

Results and conclusions
The computerized model ONIMIL (acronym for onion downy mildew) runs with a time step of 12 h (8 am-8 pm and 8 pm-8 am). It uses hourly values of air temperature, relative humidity and rain to define the day when sporulation occurs, the sporulation rate, the survival rate of spores, the day when infections are established, the infection rate, and the period when primary symptoms appear [1]. Validation showed that ONIMIL was able to forecast well downy mildew onset [2]: no infection was observed without model signalling, but in some cases, early in the onion-growing season, the presence of weather conditions favourable to infection, singled out by the model, did not determine infection establishment in the field. Model simulations were more accurate when they were linked with information on both the presence of primary inoculum and the growth stage of host plants. The latter stage being not susceptible to infection before stage F of Rey's scale [3], when the 1st leaf gets dry and the 5-7th leaves appear [4]. A warning system for primary infections of downy mildew on spring-seeded onion crops should be based on three steps: (i) monitoring host growth by means of field surveys or by using growth models; (ii) monitoring the presence of primary inoculum by means of spore traps or by looking at either autumn-seeded or seed-mother crops; (iii) using the ONIMIL model. Warnings should be made when (i) onion plants reach a growth stage susceptible to infection; (ii) the primary inoculum is present; and (iii) the model points out that weather conditions are favourable to infection establishment.

1. Battilani P, Rossi V, Racca P, Giosub S, 1996. EPPO Bulletin 26, 567-576.
2. Battilani P, Rossi V, Racca P, Giosub S, Roberti R, Flori P, 1996. Proceedings Workshop on Decision Support Systems in Crop Protection: SP Report No.15: Danish Institute of Plant and Soil Science, pp. 9-8.
3. Rey C, Stahl J, Antonin P, Neury G, 1974. Revue Suisse de Viticulture, Arboriculture, Horticulture 6, 1-104.
4. Battilani P, Bottazzi R, Racca P, Rossi V, 1998. Atti Giornate Fitopatologiche, in press.