Assessing the risk of primary infection of cereals by barley yellow dwarf virus in autumn in the Rennes basin of France

SUMMARY In the Rennes basin, Rhopalosiphum padi is anholocyclic and represents more than 90% of suction trap catches of potential vectors of barley yellow dwarf virus (BYDV) during autumn. From 1983 to 1987 the possibility of predicting the risk of BYDV infection of batches of barley test seedlings (sampling units) exposed each week from September to December near a 12.2 m high suction trap was investigated. Three kinds of variables were checked as possible predictors: weekly mean or maximum temperatures; weekly catches of R. padi (including or excluding males); and percentage of sampling units infested by aphids. Three contrasting examples were observed: during the first three years (1983–1985), infection was high and its change with time followed temperature, aphid catches and plant infestation changes; in 1986, high numbers of aphids caught and a high proportion of plants infested resulted in only low infection and in 1987, both infestation and infection were very low. Simple linear regression analysis showed that the more reliable predictors of infection were the proportion of infested plants and to a lesser extent the numbers of trapped aphids. Multiple linear regressions including either of the three groups of ‘predicting’ variables did not result in any improvement in the prediction. At a practical level, the use of counts of aphid catches would seem a better compromise between accuracy and consistency of prediction and ease of gathering data than that of plant infestation but any significant improvement of the prediction should be sought in an early estimate of the amount of virus available to aphids before they colonise the plants.