Richard Harrington, Department of Entomology and Nematology, IACR-Rothamsted, Harpenden, Herts, AL5 2JQ, UK; Garth Foster, SAC Environmental Division, Auchincruive, Ayr, KA6 5HW, UK; Steve Tones, ADAS, Mamhead Castle, Mamhead, Exeter, Devon, EX6 8HD, UK; Ian Barker, CSL, Sand Hutton, York, YO41 1LZ, UK; and Derek Morgan, CSL, Sand Hutton, York, YO41 1LZ, UK
The epidemiology of BYDV in autumn-sown crops in the UK will be outlined. Research on factors affecting primary infection and secondary spread of the disease will be summarised. A weather-driven regional stochastic simulation model based on this research will be described together with the results of a small-scale model validation exercise. The influence of field characteristics on incidence will be shown. Progress with integrating the model within a decision support system and prospects for implementation of the system will be discussed. Gaps in knowledge will be highlighted.
Epidemiology of BYDV in the UK
Rhopalosiphum padi and Sitobion avenae are the main vectors. Rhopalosiphum maidis is sometimes present. PAV is the most common isolate. MAV and RPV are also present. R. padi is largely holocyclic in the UK, but significant numbers of individuals from anholocyclic clones are always present and colonise cereals. S. avenae is largely anholocyclic in the UK. R. maidis is entirely anholocyclic throughout its range. Virus enters crops as a result of winged aphids flying in from reservoir hosts, which comprise many species of the Poaceae. Direct transfer from previous crops or volunteers may also occur but this is not considered here. Spread of the virus is a result of offspring of the colonisers moving through the crop when conditions permit.
Factors affecting primary infection
In autumns following mild winters and wet summers the number of cereal-colonising aphids is greater than in autumns following hard winters. Mild winters improve the chances of survival of individuals from anholocyclic clones and this is reflected in the population structure the following autumn. Wet summers improve survival of reservoir host plants that bridge the gap between crops. Thus vague early warnings are possible, but the modelling of infection begins with monitoring colonising aphids. Some colonisers fly within crops, increasing the potential number of virus foci.
Factors affecting secondary spread
Temperature is the main determinant of secondary spread through its influence on aphid development, reproduction, mortality, movement and efficiency of virus acquisition and inoculation. It also affects the latent period of the virus in the aphid and the crop. Rain and wind affect aphid movement and mortality. The role of aphid natural enemies in BYDV spread is unclear.
Modelling virus spread
The model is initiated on the basis of suction trap catches of aphids. These are adjusted to take account of the proportion of the sample expected to be from anholocyclic clones, the proportion of these expected to be carrying BYDV and the average number of flights made within a crop in previous trials. Trap samples are converted to numbers per unit area of crop on the basis of an established relationship with numbers on sticky wire traps sited over a crop. Numbers per plant are then calculated on the basis of planting density. The model unit is a single plant. The model is temperature driven and aphids grow, reproduce, move and die on the basis of algorithms developed from experimentation within this project or from the literature. The effects of temperature on acquisition and inoculation efficiencies and latent periods are also accounted for. Rain and wind are not yet included as driving variables. Output is currently in the form of percentage plants infected. Yield and economic data are not yet incorporated. A sensitivity analysis identified virus latent period in the plant, low temperature aphid mortality, the number of infectious winged aphid immigrants and the dispersal rate of wingless aphids as the most critical factors.
The model output for aphid and virus incidence was tested against independent data collected from small plots at three contrasting sites over two years for two of the sites and one year for the other. Subplots were sprayed at different times during the autumn and winter to halt further spread of BYDV, and BYDV incidence was assessed in spring. Aphid populations were simulated well in three of the five trials, but were lower than predicted in the other two. Final virus incidence was predicted well in the same three trials but its progress curve was not always predicted accurately.
Over three growing seasons, 623 unsprayed cereal crops were surveyed in autumn for aphid abundance and in spring for BYDV incidence. Values for forty five categories of field characteristic were recorded and a multivariate analysis used to assess their relationships with aphids and BYDV. Aphid and virus incidences were strongly correlated. There was more BYDV (P<0.01) in earlier sown crops, crops closer to the sea, crops around which arable land was less dominant, and in east (MAV) and south west (PAV) facing crops.
Development of a decision support system
The model will run under an umbrella decision support system being developed independently and there has been close liaison between the modeller and the DSS team. The model needs to be greatly simplified in order to run fast enough within the DSS.
Gaps in knowledge
Further quantification of aphid winter mortality is needed. There remain difficulties in monitoring colonising S. avenae.
Prospects for implementation
The next stage is to test the model on a commercial scale. If this is successful it may be in use in three year's time