The interactions between plants and associated insects are complex and often mediated by volatile organic chemicals (VOCs). Infection by viruses alter plant volatile signatures and consequently affect how insects interact with plants. In previous flight arena experiments in the Carr lab, cucumber mosaic virus (CMV) was shown to cause plants to emit volatile chemicals that attract bumblebees (Bombus terrestris). Bumblebees are important tomato pollinators but do not transmit CMV. For wild plants under natural conditions, helping host reproduction by encouraging bee visitation, pollination and increased host reproduction might represent a payback by the virus to susceptible hosts. Fieldwork at RHS Garden Wisley was used to test this hypothesis for the first time in tomato (Solanum lycopersicum) under field conditions by comparing pollination rates and the resulting seed production of CMV-infected and mock-inoculated tomato plants.

In the 2018 field season, after growth and inoculation of tomato plants (either with CMV or sterile water for mock-inoculation) the plants were placed outside in twelve plots each plot consisting of twelve plants of either CMV-infected or mock-inoculated plants. Each plot was split in half with 6 plants in each and one half of each plot was covered with insect-proof netting. At the end of the growing season over 1,550 tomato fruits were harvested. These were left to rot slightly so as to loosen the flesh and then their seeds were extracted and washed by hand. These seeds were then counted (with help from an image processing system) so as to be able to compare the reproductive success for each plant and even to the resolution of each flower. In total, 93,948 seeds were extracted and counted. Mock-inoculated plants were found to have produced similar numbers of seeds whether or not pollinators were excluded. CMV-infected plants however produced a greater number of seeds (if their size was taken into account) when pollinators were allowed access to flowers. In the 2019 season the same plot structure was used but plants were not covered with insect-proof netting. Bumblebee visits to these plants were recorded during day-long bumblebee counts. A binomial model was fitted to the bumblebees first choices and a geometric model to their subsequent choices, both times using a Bayesian likelihood-based technique. This modelling showed that bumblebees were not visiting CMV-infected flowers more frequently and in fact had a preference towards flowers from mock-inoculated plants.

It therefore appears that virus-induced changes in bee perceivable VOC emission cannot solely explain the observed rescue of seed yield. I also report on observational studies on the experimental plots comparing the effects of CMV-infection status on the incidence of vector (aphids) and non-vector (a leaf-mining fly) herbivores and a non-viral disease of tomato (late blight disease).