| Abstrakt: |
Interaction networks between plants and their floral visitors are often studied with a plant-centered approach, and the results obtained with these data may generate incomplete conclusions on network topology. Thus, we propose to elucidate how biases associated with the approach to data collection can influence the topology of interaction networks by examining species richness, interactions, generalization levels, and nestedness in three interaction matrices: plant-centered, animal-centered, and their combination. The results of the combined data indicate an increase in connectance in relation to animal-centered networks and differences in the mean degree for plants and animals, showing the importance of complementary methods of collecting data to reduce the sampling biases of each isolated method. The nestedness level was most time significant and higher in the animal-centered networks. Our work shows that the apparently specialist behavior of some visitor species may be in part circumvented with the supplementation of plant-centered data with animal-centered data, thereby leading to a result closer to that expected by theory, which predicts a more generalist behavior. [ABSTRACT FROM AUTHOR] |
Full text from SpringerLink