The effect of habitat fragmentation on plant pollinator mutualisms: An experimental approach using wild radish, Raphanus sativus L. (Brassicaceae) as a model organism

Given the significance of habitat fragmentation to conservation of populations and species, surprisingly few researchers have purposely created an experimentally fragmented landscape and then assessed its ecological consequences. Therefore, a large-scale experiment was carried out in order to address how plant populations and their pollinator mutualists are affected by the changes in spatial distribution that accompany habitat fragmentation. Genetically distinct population of wild radish, Raphanus sativus L. (Brassicaceae), that varied in size and degree of isolation were placed in the field so that they could be pollinated naturally. Pollinator diversity, abundance, and behavior were observed and compared to fruit and seed set among the experimental plant populations. Additionally, the paternity of a sample of 14513 seeds was determined to establish the extent and pattern of gene flow along the fragmentation gradient. These data were also used to test FST, a measure of genetic distance that had been used to estimate gene flow in thousands of studies, but has never been explicitly tested.; Results indicate that isolation distance from neighboring populations diminished pollinator abundance and visitation, but plant population size had little effect on pollinators. Decreased size and isolation of plant populations resulted in decreased seed set. Much of the variation in reproduction could be explained by the changes in diversity and abundance of pollinators over time. Gene flow was highest in populations that had the most access to more populations, but population size did not affect gene flow. However, the patterns of gene flow observed were the result of patterns of pollinator movement among populations, particularly honey bees and solitary bees. Thus, habitat fragmentation appeared to affect plant-pollinator interactions by limiting the quantity and quality of pollen available to fertilize ovules. Finally, when the underlying assumptions of models of population structure were controlled, all predictions of population parameters based on genetic and geographic structure were quite accurate. Therefore, when direct estimation of pollen movement is impossible, FST remains a useful estimate of gene flow.