The relationship between aggression level and fluctuating phase of population density of root voles, Microtus oeconomus, was studied under different predation and food treatments, predator-access and food supplemented (+P,+F), predator-excluded and food supplemented (-P,+F), predator-excluded and non-supplemented (-P, -F), and predator-access and non-supplemented (+P, -F). The aggressive behaviors observed in the study were threat, attack, chase, wrestle and retreat. The sum of the accounts of three aggressive behaviors, threat, attack and chase, was defined as aggressive account, which stood for aggression level of the population. The individuals that showed active threat, attack, and chase behavior were defined as aggressive individuals.
The results showed that the proportions of aggressive individuals were variable during different fluctuating phases in the four populations. The proportions of aggressive individuals during increase and peak phases were significantly higher than those in decline phase under all treatments except that of females under -P, -F treatment. During the whole fluctuating periods, the average counts of threat, attack and wrestle in every trapping session under all treatments were all significantly positively correlated with population density. In general, experimental data from all root vole populations were consistent with what was predicted by Chitty’s polymorphic behavior hypothesis. The experimental results testified our specific hypothesis: aggressive behavior selection, which was mediated by the interaction of predation and food availability as external factors, was the main internal regulating factor in seasonal fluctuating populations of arvicoline rodents.