Interacting virus abundance and transmission intensity underlie tomato spotted wilt virus incidence: an example weather-based model for cultivated tobacco

Abstract

Through a modeling approach, we investigated weather factors that affect the summer incidence of Tomato spotted wilt virus (TSWV), a virus vectored exclusively by thrips, in cultivated tobacco. Aspects of thrips and plant biology that affect disease spread were treated as functions of weather, leading to a model of disease incidence informed by thrips and plant biology, and dependent on weather input variables. We found that disease incidence during the summer was influenced by weather affecting thrips activity during the preceding year, especially during a time when thrips transmit TSWV to and from the plant hosts that constitute the virus' natural reservoir. We identified an interaction between spring precipitation and earlier weather affecting thrips, relating this to virus abundance and transmission intensity as interacting factors affecting disease incidence. Throughout, weather is the basic driver of epidemiology in the system, and our findings allowed us to detect associations between atypically high- or low-incidence years and the local climatic deviations from normal weather patterns, brought about by El Niño Southern Oscillation transitions.

Figure 1. Simplified schematic of the disease system’s annual cycle, showing directionality of transmission events in time.

Black arrows indicate transmission events that result in TSWV’s continuous presence in the landscape. The single red arrow indicates transmission from winter annuals to tobacco in early spring; note that transmission from tobacco to other plants does not occur to influence the cycle. Tobacco is shown to be present in spring and summer, and absent during fall and winter.

Figure 2. Transmission event timeline showing the occurrence of different host groups (summer and winter annual weeds, and crop tobacco) in time, and an interpretation of how and when relevant weather parameters influence the disease system.

Note crops in TSWV susceptible crop hosts in the tobacco production regions are harvested and removed from the agroecosystem before winter annuals germinate.

Figure 3. Prediction plot for best-fit model, a regression of observed on fitted TSWV incidence, R² = 0.899.

Figure 4. Interaction effect in observed data, showing change in influence of prior-year thrips variable across range of March precipitation, and to lesser extent, vice versa.

Figure 5. Interaction effect in fitted data, demonstrating that the best-fit model accounts for the interaction.

Figure 6. Panel of diagnostic plots for best-fit model, showing conditional Studentized residuals.