Rodent and predator population dynamics in an eruptive system

A computer model of the population dynamics of introduced house mice (Mus muscidus L.), ship rats (Rattus rattus L.) and stoats (Mustela erminea L.) in New Zealand forest was constructed, to test the relative importance of food availability and predation in shaping observed small-mammal population dynamics. Ship rats and mice are the two common rodent species present in most New Zealand forests, and exhibit eruptive population dynamics. Stoats are the only common mammalian predator, and undergo large density fluctuations following periodic rodent eruptions. A number of outputs and predictions from the model were developed. The model highlights the overall importance of variation in food availability in determining the timing and amplitude of rodent population eruptions. It indicates that predators can not prevent a prey-species eruption, primarily due to differences in reproductive biology. Predation however, can delay the start of the prey-population increase during the eruption. The role of predators in limiting the peak prey-population size will depend on the size of the energy input. In a full-scale eruption following maximal tree seeding, predators cannot significantly truncate peak prey-population size. Predators should be able to significantly hasten the rate of decline in the prey populations, although the strength of predator limitation will depend on the severity of food limitation and cold-induced mortality over the same period. Predators can limit prey populations during the post-crash low phase. As with the crash phase, the strength of predator limitation in the low phase will depend on the severity of food limitation and natural mortality. The model highlights gaps in current knowledge of predator and prey species biology and ecology. The model highlights key areas where further field study should provide a better understanding of the factors driving small-mammal communities in New Zealand. (C) 2001 Elsevier Science B.V. All rights reserved.