Epidemiological theory has shown that, when host populations are at high densities, the introduction of even a small number of infected individuals can lead to a severe epidemic. This result is likely to be important in the use of species-specific baculoviruses as insecticidal sprays to control insect pests. However most such virus spray efforts assume that baculoviruses are no different from conventional insecticides. Here we use mechanistic, epidemiological models to understand the use of a baculovirus to control the Douglas-fir tussock moth. The USDA Forest Service has attempted to use the insecticidal formulation of this virus, known as "Tussock moth Biocontrol-1" (TMB-1), to reduce tussock moth defoliation, but repeated efforts have encountered a surprising problem. That is, virus epidemics occurred in both spray plots and unsprayed control plots, even though initial infection rates in control plots were very low, and even though control and spray plots were too far apart to permit spray drift. Epidemiological theory then suggests that host densities were sufficiently high in control plots to permit epidemics to occur, but an alternative hypothesis is that naturally occurring virus is far more infectious than previously believed. To test these hypotheses, we used a field experiment to estimate the infectiousness of naturally occurring and sprayed virus, and we used statistical model selection to choose between models that (1) assumed that control plots and spray plots differed only in initial densities, or (2) assumed that control plots additionally had more infectious virus.
Results/Conclusions
Our results support the hypothesis that virus epidemics occurred in both spray and control plots simply because the initial density of uninfected insects in the control plots was sufficient to lead to a severe epidemic, even though initial infection rates in control plots were vastly lower than in spray plots. Our results therefore suggest decisions to use baculoviruses in microbial control should be based on thorough sampling to accurately estimate initial densities of infected and uninfected insects. In an effort to assist decision makers, we show how the epidemic sizes change with initial densities of infected and uninfected hosts, which can be directly translated into costs of spraying versus not spraying. Our work demonstrates that epidemiological theory can be directly useful in guiding biological control programs.