Abstract
How connectivity affects plant assemblages is a central issue in landscape ecology. So far, empirical studies have produced contradictory results, possibly because studies: (1) inaccurately assess connectivity by prioritizing the respective effect of the type of habitat on plant assemblages and (2) omit the range of possible plant responses to connectivity depending on dispersal vectors. We focused on three dominant habitat types in agricultural landscapes (woodland, grassland and cropland), and analysed the effect of connectivity on herbaceous plant assemblage similarity for three primary dispersal modes (animal-dispersed, wind-dispersed and unassisted). Using circuit theory, we measured connectivity provided by woodland, grassland and cropland habitats independently. The similarity of plant assemblages was evaluated relative to the random expectation based on the regional pool. Overall, plant assemblage similarity in woodlands and temporary grasslands was dependent on connectivity, but not in wheat croplands. Only animal-dispersed species responded to connectivity. The similarity of animal-dispersed assemblages in woodlands was increased by the connectivity provided by woodland habitats, but was reduced by cropland habitats, whereas in temporary grasslands, similarity was increased by the connectivity provided by cropland habitats. Our results suggest that animal-dispersed species supplement their dispersal pathways, thus improving our knowledge of plant assembly rules in fragmented landscapes.