Assessment of source-sink theory for predicting demographic rates among habitats that exhibit temporal changes in quality

Abstract

Within the framework of Pulliam's source-sink model we tested predictions of habitat-specific demography on a "closed" population of muskrats (Ondatra zibethicus L.) occupying three contiguous habitats that exhibited temporal changes in quality and quantity. We were able to distinguish between dispersal among habitats and mortality within each habitat, and induce temporal variation in operational-habitat availability and suitability by manipulating water level. Temporal variation in population size and density among habitats supported the source-sink model and was primarily associated with habitat-specific survival rates. For example, the mean annual over-winter mortality rate of individuals in the principal source habitat (0.87) was less than in the sink habitat (0.94), and subsequently the mean annual finite rate of increase (λ) was positive in the source habitat (λ = 1.41) and negative in the sink habitat (λ = 0.90). The high recruitment rate of juveniles in the prime habitat during autumn was also associated with significant emigration of juveniles from this habitat. Emigration of adults among habitats provided support for the role of spacing behaviour in regulating the breeding density of muskrats in prime habitat. However, in contrast to the assumption of source-sink theory, year-to-year variation in survival rate in the more marginal habitats appeared to be explained more by temporal changes in habitat suitability than by density. Significant emigration of juveniles from the sink habitat was not predicted and was largely dependent on current habitat conditions. In addition, the mean annual emigration rate was lowest in the principal source habitat (0.30) and highest in the more marginal habitats (0.62). Failure to detect directional emigration from prime to marginal habitats in the spring, as predicted by the source-sink model, was likely due to declining local population size. In environments where spatial differences in habitat quality are not static, and annual change in local population size is largely independent of density, current source-sink models must be modified to better predict individual dispersal strategies.