The island syndrome in rodent populations.

Populations of rodents isolated on islands often show systematic differences in demography, reproduction, behavior, and morphology when compared to mainland populations. These differences, termed the island syndrome, include higher and more stable densities, better survival, increased body mass, and reduced aggressiveness, reproductive output, and dispersal. We synthesize information in the literature on island rodent populations and construct a conceptual model to explain the island syndrome. Population density and other manifestations of the island syndrome are predicted to increase with island isolation and to decrease with island area. The effect of isolation is direct by limiting dispersal, while the effect of area is less direct. As area increases, predators, competitors, and habitat structure increase in diversity. We suggest that the intensity or absence of density-depressing factors (primarily predation in communities comprised of only a few morphologically and trophically divergent species) is primarily responsible for the area effect and serves as a principal factor differentiating island from mainland populations of rodents. Other characteristics associated with the island syndrome may be the result of both short-term and long-term processes or responses. Short-term responses include reproductive, body size, and behavioral changes that may be phenotypic reaction norms in response to higher island densities. Thus, higher densities lead to reduced reproductive output, which then leads to greater body size. Initial behavioral differences may be owing to better survival and reduced dispersal, which result in less population turnover, greater neighbor familiarity, and less aggression. Long-term changes may be due to directional selection for reduced dispersal, increased body size in response to increased intraspecific competition, reduced reproductive output (smaller litters and delayed maturation) in response to reduced mortality schedules, and reduced aggressiveness. While the model is specific to rodent populations, the conclusions may be applicable to other systems if characteristics such as body size, vagility, and community composition are considered.