Background/Question/Methods Climate change is influencing abiotic conditions, causing plant species to experience abiotic stress, such as drought. Many insect host-pathogen interactions occur in the context of tritrophic interactions, in which the plant is the food supply of the insect host and the insect provides resources for the pathogen. In such situations, changes to plant quality or quantity could elicit changes in the host-pathogen interaction. We explored how water to Passiflora caerulea, the larval food plant of the Gulf Fritillary butterfly (Agraulis vanillae), influences the dynamics between Agraulis and its host-specific baculovirus (AgvaNPV). We manipulated water availability to 40 potted Passiflora individuals, exposing plants to high or low water availability over six months. Replicated experimental epidemics were conducted in the field by exposing each caged plant to one of four treatments: zero larvae (control), ten uninfected larvae, low-prevalence of infected larvae (10%), or high-prevalence of infected larvae (30%). Data on larval densities and fate (viral infection or pupation) was collected daily until all larvae had pupated or died. Aboveground biomass was calculated on all plants. We then used the observed epidemic trajectories to fit a stochastic Susceptible-Exposed-Infected-Removed (SEIR) ordinary differential equation model of the host-pathogen interaction using Approximate Bayesian Computation (ABC) methods. Results/Conclusions Although we detected strong effects of the water treatments on plant biomass, preliminary results show no effect on the pathogen dynamics, as measured by the fitted model parameters such as the transmission rate and environmental decay rate of AgvaNPV. Aboveground fresh biomass decreased by 50% in the low water treatment compared to the high water treatment, and aboveground dry biomass exhibited a similar pattern. Since a change in plant biomass was detected, it is surprising that no effects were observed on the disease dynamics, as virus transmission occurs by consumption of contaminated leaf tissue. Thus, the effective viral density per g leaf on high-water plants (mean±SE: 0.00306 ± 0.00009 cadavers/g leaf) was approximately double that of the low-water plants (mean±SE:0.00578 ± 0.00049 cadavers/g leaf), and yet viral transmission was unchanged. One potential mechanism is that leaf chemistry changed and counteracted the effects that water availability had on biomass. Follow-up laboratory experiments will be presented and used to determine which mechanisms were at play. Our results suggest that plant drought stress could affect host-pathogen dynamics in opposing ways via direct versus trait-based tritrophic effects, leading to complicated and non-linear responses in this tritrophic interaction.
