1. Antagonistic interactions in herbivorous insects are often density-dependent, so rates are predicted to vary dynamically over time as density changes. Fatal intraspecific interactions, especially cannibalism, occur between equal-aged larvae in young first- and second-instar Spodoptera frugiperda (J.E. Smith). 2. A cannibalism experiment was conducted, starting with seven different densities of neonate S. frugiperda larvae, each replicated 50 times. Larvae were examined daily for the duration of the first and second instars (7days). Seven-day mortality was density-dependent. 3. A stochastic mathematical model was developed in which per-capita mortality from antagonistic interactions among equal-aged larvae varies dynamically as density changes. A maximum likelihood method was developed to estimate the conditional per-capita mortality rate from antagonistic interactions given an intraspecific encounter. An alternative model with mean-mortality from antagonistic interactions that depends only on the initial larval density was also developed. 4. The models were fitted to the experimental data, and compared using log-likelihood. The dynamic model fitted the cannibalism data significantly better than the time-averaged mortality model for all starting densities for the experimental data, implying that density-dependent mortality varied dynamically over time even within short 7-day periods. 5. The conditional per-capita mortality rate from antagonistic interactions was also density-dependent, possibly because encounters became more aggregated at higher density, or because the probability that a larva died from the interaction was higher at higher density, or both.
- Fall armyworm
- Herbivore cannibalism
- Maximum likelihood
- Stochastic mathematical model