Using agent-based models to determine how Gypsy moth feeding behavior affects transmission of LdNPV

Background/Question/Methods

The gypsy moth, Lymantria dispar, is an economically important defoliator of hardwood trees. One method of controlling this pest uses the naturally-occurring baculovirus LdNPV, which kills and liquefies larvae. To effectively use this virus as a biocontrol agent, it is necessary to know how likely it is that gypsy moths might develop resistance to it. For disease resistance to spread, individuals must vary in their susceptibility to infection and this variability must be heritable. While it is known that risk of LdNPV infection varies among individual gypsy moths, little is understood about the factors causing this variation.

 My aim is to determine if gypsy moth feeding behavior is a factor that affects risk of LdNPV infection. To determine if feeding behavior affects probability of infection, I conducted an experiment where individual larvae fed on leaves infected with LdNPV-infectious cadavers. For all individuals, I took pictures of the leaves before and after feeding, and recorded the total leaf area eaten, the closest distance eaten to a cadaver, and the infection outcome. I then constructed a mathematical model to evaluate the probability of infection given individual feeding behavior. Using this information, I created an agent-based simulation in order to assess whether differences in larval feeding behavior affect risk of transmission.

Results/Conclusions

The data show that the percentage of larvae dying of LdNPV infection decreases as the distance eaten to a cadaver increases (logistic regression, p = .04). The total amount of leaf area eaten does not, by itself, affect risk of infection (p = .08). However, the best fit mathematical model indicates that the probability of infection is determined by both the closest distance eaten to a cadaver and the amount of leaf area eaten near it. Preliminary results from the agent-based model suggest that changes in feeding behavior dramatically alter an individual’s risk of LdNPV infection.

These results indicate that feeding behavior has the potential to play a role in gypsy moth-LdNPV dynamics. The future direction of this project is to create a larger agent based model to simulate a population of larvae feeding in an environment containing LdNPV-infectious cadavers. In this simulation, feeding behavior is assumed to be heritable, and risk of infection is determined by an individual’s feeding history. The simulation will run for successive generations and will show if changes in larval feeding behavior can affect the evolution of LdNPV resistance in gypsy moths.