SEARCH

SEARCH BY CITATION

Changes in the global environment are modifying the geographical locations of habitats suitable for plant growth. The capacity of plants to migrate to sites of suitable environmental quality will strongly influence future distributions of plant diversity. However, it is not well understood how rates of plant migration are influenced by the habitat loss and habitat fragmentation that characterise contemporary landscapes. In this study we develop a model that can predict migration rates in both intact landscapes (potential migration rate) and in fragmented landscapes (realised migration rates). Migration rates in fragmented landscapes might be slower for many reasons. In this study we focus on two, non-exclusive reasons. First, the processes that move seeds may break down in fragmented landscapes causing seeds to be dispersed shorter distances. Second, in fragmented landscapes some proportion of seeds will not be deposited in habitats suitable for recruitment. We describe the breakdown of dispersal processes as a competing risk between the factors influencing dispersal in intact landscapes and the factors that may disrupt dispersal processes in fragmented landscapes. We show how the parameters that influence dispersal in fragmented landscapes can be estimated, and how these estimates can be used to forecast migration rates using an integrodifference equation (IDE). The forecasts of the IDE described the effects of reduced dispersal distances adequately. However, the IDE produced biased estimates of the effects of a reduction in plant habitat on migration rates. Model analyses showed that, although we can expect realised migration rates to be lower than potential migration rates, we can also expect the sensitivity of migration rate to habitat loss to vary. In addition, simulations showed that the qualitative nature of the responses of migration rate to habitat loss were variable – some model species responded non-linearly to habitat loss, others responded linearly. While our method provides guidelines for empirical data collection and model parameterisation, we recognise that obtaining these data will be challenging.