Aphids cause large losses to UK agriculture through direct feeding damage and by transmitting a range of diseases. A better understanding of how aphid populations are affected by the weather will allow us to give forewarning to farmers. This should allow them to prevent damage while minimising their use of insecticides.
Data on aphid activity are available from a number of locations where suction traps have been operated for up to half a century.
Simple analyses of these data have suggested a strong link between aphid activity and weather, in particular the temperature. This link is currently used to predict the earliness of the arrival of aphids based on the mean temperature during certain pre-specified months: lower temperatures early in the year are considered to delay the timing of the first aphid flights.
However, it is unlikely that such simple methodologies are getting the most out of the data. The actual relationship between climate and aphids is unlikely to be so simple, and first arrival itself is not necessarily the most important metric in determining levels of disease transmission.
BioSS is undertaking a research project to devise and test new techniques for modelling aphid activity in the context of climate data. We are investigating several avenues.
Leveraging past work in phenology, more complex relationships between temperature and aphid populations are being considered. This includes spline models, which allow temperature to have a smoothly varying effect on aphid timings, and mechanistic models, which try to produce ecologically meaningful explanations of how changing temperatures might trigger aphid flights.
We are considering other metrics of aphid activity that may be both more reliable and more relevant to agricultural users. For example, quantile days may be utilised. Alternatively, we have developed novel methods to relate temperatures to the entire distribution of aphids throughout the year. Thus, instead of giving just the date of arrival, we can attempt to predict the intensity of aphid activity on any particular day, and possibly involve aphid counts (including at different sites) already made during that year to update these predictions.
Finally, our exploratory analysis has showed the relevance of wind to aphid counts. By developing an understanding on how these factors can affect aphid behaviour or trap efficiency, BioSS can incorporate these factors into its models and obtain better predictions.
We seek to continue to improve our models, and also to apply them to data from other sites, with the potential to combine data from multiple locations.