I spent Tuesday morning in a really well-organized symposium entitled “Mutualistic Networks”. Headed up by Jordi Bascompte, the collected talks focused on the network architecture of mutualistic interactions, mostly among plants and their various insect pollinators. I came in with only a very basic understanding of matrix-based interaction networks, but Bascompte’s introduction to the session provided me with enough background to get up to speed. Originally coevolution was conceived of as occurring in a geographical mosaic, a concept first proposed by John N. Thompson. More recently the concept of a mutualistic network has been shown to explain how these systems can be comprised of a vast array of interconnected mutualisms, some specialist and some generalist. Theoretical investigations of these networks has shown that the network architecture is key to stability: many networks are unstable, but networks with a “nested” structure are stable. Nested networks are defined by their asymmetric specialization. In the case of plants and their pollinators, this means that the majority of plant species which only have one pollinator (i.e. specialists in their pollinator requirement) are paired with an insect species capable of pollinating numerous plants species (i.e. generalist pollinators). The opposite is also true: the majority of plants that have many pollinators (i.e. generalists in their pollinator requirement) are paired with insects that rely on that plant for provision (i.e. specialist pollinators). The obligate mutualisms that are often used to illustrate the concept are rare in nested networks, as they involve paired specialists.
Why this pattern would be most stable is pretty intuitive. Plant species that require a particular, specialized insect will be more likely to be pollinated if that insect’s population is boosted by being provisioned by multiple plant species. But plant species that don’t require a particular insect can rely on a series of relatively rare pollinators who specialize on that plant alone. Insects that require a particular plant species will be more likely to find food if that plant’s population is boosted by being pollinated by multiple insect species. But insect species that don’t require a particular plant can rely on a series of relatively rare plant species requiring that particular insect’s pollination services.
It is always nice when theory suggests how nature should look, but do observed mutualistic networks really display nestedness? Jens Olesen presented work on butterflies and their nectar plants suggesting that in accordance with the nested network architecture the generalists have the highest abundance. His work also highlighted the dynamic nature of these networks: while generalists maintained a steady presence throughout the period of observation, specialists blinked in and out, “re-wiring” the network. Seventy-eight percent of links changed during the study, which is a pretty astounding amount of turnover. Pedro Jordano showed how the phylogenetic relationships between plants and between their pollinators maps onto the nested network architecture, with plant phylogeny being the best predictor of network structure. Victor Rico-Gray presented some interesting parallels in ants showing that smaller networks were often not nested but larger networks were, suggesting that the stability of larger networks is particularly sensitive to network architecture. He also introduced the concept of “intimacy”, the degree to which mutualistic pairs are symbiotically paired in space and time. Very intimate species are likely to be obligate, leading to a compartmentalized structure (it doesn’t get much more compartmentalized than a series of obligate pairs!). Where there is less intimacy there is more generalism and a more nested architecture. Rico-Gray also showed that while the networks change over time due to changes in linkage and species turnover, the nested architecture is consistent. In a large data set comprising seven years of observation, Theodora Petanidou showed similar results, with dramatic dynamics in the network linkages but relatively little change in the nestedness of the architecture. Jane Memmot rounded out the experimentalists with an interesting talk describing the necessity of restoring mutualistic networks if you want to restore a degraded ecosystem. Overall the empirical results shown by these speakers was impressive, although several audience members did question whether inconsistencies in sampling effort rather than actual community composition accounted for the observed dynamics through time. Really, it is hard to believe that a community could be so robust to such large fluctuations in interrelationship.
One of the things that Bascompte’s theoretical work uncovered was what I would call “apparent mutualism”. Because most plants were generalist in their pollinator requirements, many insects who relied exclusively on that plant for resources end up competing. But Bascompte discovered that while the pollinators competed to gather resources from the plants, the abundance of the plants was enhanced by enjoying multiple pollinators, which reduced competition between its pollinators. Colin Fontaine’s talk contrasted the network architecture of antagonistic networks (involving predation or parasitism) with the nested structure of mutualistic networks. When interactions are exploitative, the stable network architecture is “modular” involving small groups of relatively specialist consumers and victims. Why wouldn’t there be more generalists in an antagonistic network? Well, from the victim perspective it is clear that it would not be good to be “generalist” in your vulnerability to consumers, even if most of those consumers were specialists. Similarly, a predator that is a generalist on many prey is likely to be so successful that it over-consumes all of its prey.
I was really impressed with the amount of insight that paired theoretical and experimental work has produced in this area. During the discussion period I brought up the question of what kind of evolution was required in order to explain the consistent nested pattern of mutualistic networks. We can see that these networks are in a very stable configuration, but what allowed them to get to that configuration? For me the fundamental question here is whether selection acting on individuals of each species is sufficient to explain their evolution to stable configurations. Bascompte’s “apparent mutualism” suggests that benefits might flow not only between plant and pollinator, but also between pollinators sharing common plants and plants sharing common pollinators. Does that mean that individual selection would provide a sufficient explanation of the network architecture observed in nature?
I was able to attend this meeting thanks to funding from the Pratt Institute Mellon Fund for Faculty Travel.