Temporal dimension in plant-pollinator networks

Submitted by editor on 5 April 2016.Get the paper!

Network theory is widely used to explore the structure of ecological interactions. Consequently, network tools have permeated the ecological literature during the last decade. For example, plant-pollinator interaction networks accounted for nearly 250 papers in the last five years (Thomson-Reuter web of science). Despite this profuse literature, network theory has been seldom used to understand the dynamics and structure of ecological interactions at the individual level.

In individual-based pollination networks, plant individuals are considered nodes connected between them through pollinator visits. These connections depend on the temporal coupling between plant flowering and pollinator availability. Individual plants usually show disparate flowering schedules, causing the topology of the interaction network to change if pollinators do exhibit temporal turnover. These changes could be more intense if the plant shows a highly generalized pollination system.

In our paper (Oikos 125(4): 468-469), we present the temporal changes occurring in a pollination network established between individuals of the extremely generalist herb Erysimum mediohispanicum (Brassicaceae). We explored this dynamics in a controlled population during two flowering seasons, using metrics (modularity, clustering, and core-periphery structure) related to properties influencing the functioning of networks, such as cohesion, transitivity, and centralization.

The yellow flowers of Erysimum mediohispanicum.
Photography by F. Perfectti

Erysimum mediohispanicum is a biennial monocarpic herb endemic to the Iberian Peninsula. In southeastern Spain, this species is found in montane regions between 1100 and 2000 m of altitude, inhabiting forests and subalpine scrublands. These plants usually produce one to eight reproductive stalks displaying bright yellow flowers after. This study was conducted in an E. mediohispanicum population located in Sierra Nevada National Park (Granada, Spain). In this area, E. mediohispanicum is patchily distributed forming small populations comprising tens to several hundreds of individuals. This plant usually blooms from late May to late June and their flowers are visited by more than 200 species of insects, constituting a highly generalized pollination system.

Authors José M. Gómez and Javier Valverde (right) at the experimental population. Photography by F. Perfectti

We found that network topology varied widely over time as a consequence of the differences in plant phenology and the idiosyncratic and contextual effect of pollinators. When all temporary networks were integrated, the total network became cohesive (non modular), transitive (locally clusterized), and centralized (showing a core-periphery topology). These topologies may entail important consequences for plant reproduction, through favouring biparental inbreeding or promoting a spatial genetic structure. Finally, our study highlights the necessity of making comprehensive temporal sampling when trying to build reliable interaction networks.

This paper, a chapter of Javier Valverde’s PhD thesis, is framed in a scientific project which ultimate aim is to analyse how and why individual plants differ in their interactions with pollinators and which are their evolutionary consequences. Our initial null hypothesis was that, in pollination generalist systems, plant genotypes in a population interact with random subsets of the overall pollinator pool. However, we think that the generalization degree of a given plant could be non random, but related with some intrinsic and extrinsic factors, such as its microenvironment, its spatial location in the population, and more important, its phenotype and genotype. Plants exhibiting different values for these factors would attract different subsets of pollinators. These inter-individual differences could promote a structured pattern of interaction among the plant and its pollinators, a phenomenon that we have named "structured generalization". We propose that structured generalization occurs when there are non-random inter-individual differences in generalization, implying that the pollination and mating networks will be structured in subgroups of plants sharing similar pollinators. Our current work is focused in exploring this idea, as a way to explain why there are some many plant species with generalist pollination systems.

Javier Valverde photographing Erysimum mediohispanicum pollinators. Photography by F. Perfectti



Twitter:  @fperfectti

Insights into Oikos papers