Using metabarcoding to reveal and quantify plant-pollinator interactions

Abstract

Given the ongoing decline of both pollinators and plants, it is crucial to implement effective methods to describe complex pollination networks across time and space in a comprehensive and high-throughput way. Here we tested if metabarcoding may circumvent the limits of conventional methodologies in detecting and quantifying plant-pollinator interactions. Metabarcoding experiments on pollen DNA mixtures described a positive relationship between the amounts of DNA from focal species and the number of trnL and ITS1 sequences yielded. The study of pollen loads of insects captured in plant communities revealed that as compared to the observation of visits, metabarcoding revealed 2.5 times more plant species involved in plant-pollinator interactions. We further observed a tight positive relationship between the pollen-carrying capacities of insect taxa and the number of trnL and ITS1 sequences. The number of visits received per plant species also positively correlated to the number of their ITS1 and trnL sequences in insect pollen loads. By revealing interactions hard to observe otherwise, metabarcoding significantly enlarges the spatiotemporal observation window of pollination interactions. By providing new qualitative and quantitative information, metabarcoding holds great promise for investigating diverse facets of interactions and will provide a new perception of pollination networks as a whole.

Figure 1. Total DNA yields (ng 100 μl⁻¹, mean ± SD, n = 5) from 10,000 pollen grains of Lilium sp. Hippeastrum sp. and Chrysanthemum sp.

Figure 2. Boxplot showing the number of ITS1 or trnL sequences obtained from varied amounts of pollen DNA from Lilium sp., Hippeastrum sp. and Chrysanthemum sp. in artificial mixtures.

The crosses indicate the number of sequences obtained when the focal species was alone.

Figure 3

Quantification of plant-pollinator interactions in the Rhododendron ferrugineum communities (a) Boxplot of the number of total ITS1 and trnL sequences obtained from pollen loads of the main insect taxa. Triangles represent the mean values. One A. mellifera individual with more than 4,000,000 ITS1 sequences was not included in the figure, (b) Relationship between the number of pollen grains and the number of ITS1 and trnL sequences of R. ferrugineum across insect taxa (dashed lines represent standard deviations), (c) Relationship between the number of visits received by each plant species and the number of their ITS1 and trnL sequences in the insect pollen loads. P values were obtained with Kendall tau rank correlation tests.