Source-sink dynamics of bumblebees in rapidly changing landscapes.
Bumblebees inhabit spatially heterogeneous landscapes that are likely characterized by population sources and sinks. To date, most studies of bumblebee habitat requirements have examined static relationships between worker abundance and habitat characteristics. However, if habitat types are linked by dispersal, source-sink dynamics could alter overall population sensitivity to habitat loss, changing conclusions from static approaches. Here, we used empirically derived spatial matrix models to study source-sink population dynamics of bumblebees in heterogeneous environments. We used these models to assess population sensitivity to habitat loss and to examine the population dynamics that could occur under rapid habitat change. Loss of natural habitat reduced long-term population growth rates, and more natural habitat was required to sustain bumblebee species with long-distance dispersal than species with short-distance dispersal. The long-term relative density of colonies in each habitat type depended on landscape structure and bumblebee dispersal ability. Under most conditions, high-quality habitat contained higher densities of colonies than low-quality habitats. However, low-quality habitat could contain higher densities than high-quality habitats if populations were in long-term decline. Rapid loss or restoration of natural habitat produced transient population dynamics that differed from long-term dynamics. After large landscape perturbations that affected population structure, transient dynamics persisted for 4-8 years and were longest for species with short dispersal in highly aggregated landscapes. While transient dynamics were short in duration, they caused large effects on long-term population density. Synthesis and applications. When habitats are linked by dispersal or populations have recently experienced perturbations, patterns of abundance on the landscape may not reflect variation in habitat quality. Spatially structured matrix population models are practical tools than can be used to account for nonequilibrium and source-sink dynamics. For bumblebees, population sensitivity to habitat loss and the duration of transient dynamics depend on dispersal ability and landscape configuration. Demographic approaches such as ours can help to disentangle population patterns from processes and will therefore be valuable for guiding conservation and management.