Current infection risk prediction models utilize environmental parameters and field records, but do not take into account the estimated inoculum potential within the orchard. The object of this study was to survey the accuracy of three simple prediction methods under Hungarian climatic conditions, which could easily be used by the farmers. We also tested whether the accuracy of infection risk predictions can be improved by taking into consideration the incidence and/or rate of flower colonization by Erwinia amylovora.

After preliminary investigations in 1999-2001, data concerning the weather-related infection risk were recorded in 5 apple and 1 pear orchards in 2002, and in 12 apple and I pear orchards in 2003. The weather data were processed by the easy-to-use risk assessment models of the mean temperature prediction line (MTL), Smith's Cougarblight 98C and Billing's integrated system (BIS), and by the Maryblyt^TM 4.3 computer-assisted model for reference. The population size of E. amylovora in the flower samples was estimated within an order of magnitude by PCR.

For all years and orchards tested, Maryblyt indicated 35 days on which there was an acute infection risk. The same days were indicated by all 3 methods in 23 cases (66%), 8 days were indicated by 2 methods (23%) and 4 days were indicated by 1 method only. A similarly good correlation was found for prediction of the date of the first massive infection risk: in 2003, for instance, there was a perfectly consistent prediction by all 4 models in 9 of the 13 participating orchards. A coincidental forecast was provided by 3 of the 4 models in the other 4 orchards.

The results indicate that any of the risk assessment models could provide an increased accuracy of the actual infection risk prediction if combined with an estimation of the incidence of Erwinia amylovora colonization in the open flowers. We found no convincing differences in the size of the epiphytic population in flowers of cultivars possessing high or low susceptibility to Erwinia amylovora.

We conclude that the easy-to-use methods tested could be used by the fanners to recognize weather-related risks, especially when coupled with an estimation of the proportion of the pathogen-infested flowers. This local prediction would provide rapid information (faster than the regional forecast systems) specifically for a given orchard.