Do pollination syndromes cause modularity and predict interactions in a pollination network in tropical high-altitude grasslands?
Article first published online: 9 MAY 2011
© 2011 The Authors
Volume 121, Issue 1, pages 35–43, January 2012
How to Cite
Danieli-Silva, A., de Souza, J. M. T., Donatti, A. J., Campos, R. P., Vicente-Silva, J., Freitas, L. and Varassin, I. G. (2012), Do pollination syndromes cause modularity and predict interactions in a pollination network in tropical high-altitude grasslands?. Oikos, 121: 35–43. doi: 10.1111/j.1600-0706.2011.19089.x
- Issue published online: 21 DEC 2011
- Article first published online: 9 MAY 2011
- Paper manuscript accepted 4 April 2011
The concept of pollination syndromes has been widely questioned, since plant–pollinator interactions have proved to be more generalist than was previously thought. We examined whether the network of a tropical high-altitude grassland contained groups of plants and pollinators that interact preferentially with each other. A general binary matrix was created. To assess the robustness of myophily, in all analyses we considered: 1) the whole network, 2) the network after the wasps were removed, and 3) the network after the flies were removed. For each network we evaluated whether: 1) the observed interactions were more related to syndromes than expected by chance, compared to an expected matrix; 2) there was a modular structure; 3) the modules found were more related to syndromes than expected by chance, compared to another expected matrix; 4) the syndromes were equally robust. For this analysis, the general matrix was subdivided into smaller matrices that included each pollination syndrome separately. To test the influence of the functional groups of pollinators and the phylogeny of plants, in addition to the general matrix, we also considered the first expected matrix, a quantitative functional group and a plant phylogeny matrix. The pollination syndromes determined the pattern of interactions in the network: 69% of the total interactions resulted from the functional group of pollinators predicted by the plant syndrome. The network showed greater modularity (13 modules) than expected by chance, mostly consisting of the expected functional groups of pollinators and plant syndromes. The modules were associated with pollination syndromes more than was predicted by chance. Most of the variation in interactions was explained by functional groups of pollinators or by plant syndromes. Plant phylogeny did not account for a significant amount of variation in the interactions. Our findings support the concept of pollination syndromes. However, the interactions were not equally predicted by different pollination syndromes, and the accuracy of the prediction was strongest for ornithophily and melittophily.