Perhaps nestedness is just an artifact of ecological opportunity (and not stability)

One of my chief interests is stability: I am curious about what allows for the persistence of genes, individuals, groups, species, and communities. This is a broad question and it may not have single, simple answer, but it is exciting to think that there may be ‘rules of stability’ in nature that might help us explain existing natural phenomena or predict future evolutionary patterns. Recent research has shown that particular ecological community structures tend to me more common than others. The network of interactions between species in a community could easily be assembled at random, but these networks often show a consistent pattern, suggesting that some evolutionary process is making particular patterns of interaction more likely than others.

What might cause particular patterns of interconnectedness to emerge in networked ecological communities? Well, it turns out that there are two very different answers to this question based on two very different explanations.

The first explanation focuses on particular species and, in the end, on individuals from these particular species: an ecological community might show a particular pattern of interaction because as that community is assembled by adding new species, individuals from these new species will only be able to find ecological opportunities in particular niches. In other words, once a community has added enough species, any additional species that join the community will be constrained as to where they ‘connect’ to other species. For example, if a particular species E could prey on species A, B, or C, but all three of these species are already being preyed upon by species D, then there may not be enough ‘ecological opportunity’ to allow species E to join the community. Or, alternatively, species E might outcompete and displace species D, an outcome that would change species identity but maintain the overall structure of the community. In order for the community structure to actually change, species E has to ‘find a place’ in the network that will allow it to survive, and those places may become limited as an ecological community diversifies. These limitations could lead to consistent patterns, suggesting that the assembly rules of ecological communities are responsible for any consistency in their overall structure.

The second explanation does not depend on the ecological opportunities afforded to particular species but on the overall stability of the ecological community as a whole. Perhaps species can be added to the community, but if they are added in the ‘wrong’ place in the network of interactions they may destabilize the entire network. At minimum this destabilization would force the local extinction of some species in the network, simplifying the ecological community until it reached a more stable configuration. The more extreme possibility is that the entire network could become destabilized, leading to the local extinction of the entire community. Regardless of how extreme the destabilization effect might be, this explanation of community configuration rests not in what is possible for new species attempting to join the community but on the stability properties of the entire community as a whole.

As I have discussed before, mutualistic networks show a remarkably consistent architecture that has been described as “nested”. In a nested mutualistic network, mutualists and their hosts are not randomly paired with each other. To make my explanation easier to follow I will consider pollinators and their host plants, although the description below could apply to any community containing mutualist species. Rather than having pollinators randomly paired with host plants, what we find in a nested community is that a few ‘generalist’ pollinators service a large number of host plants and a few ‘generalist’ plants play host to a large number of pollinators. ‘Obligate mutualisms’ — in which two species rely exclusively on each other — are rare. As such, nested communities are inherently redundant, as the loss of the ‘generalist’ species is made unlikely due to its large number of partners. Because most species pair with these few generalists, the assumption is that they too are protected from going extinct, even if a single generalist partner is their only partner. Previous theoretical work (Bastolla et al. 2009, Thebault & Fontaine 2010) has demonstrated that this nested architecture makes these ecological communities more stable, suggesting that their form may be shaped by their overall stability (explanation #2 above) rather than their assembly rules (explanation #1 above).

A new paper in Nature entitled “Disentangling nestedness from models of ecological complexity” calls into question the idea that nested mutualistic networks are in fact more stable. Authored by Alex James, Jonathan W. Pitchford, and Michael J. Plank, the research bases its claims on an analysis of fifty-nine empirical plant-pollinator networks that were analyzed for their “persistence”, a measure of long-term stability. Essentially this is a model that runs within the structure provided by existing communities; by modifying the model, the authors can better investigate the properties of these networks. For instance, one manipulation that the model allowed was the switching off of mutualistic connections. Although this creates a completely artificial community, it allowed the researchers to disentangle the effects of competition from the effects of mutualism.

The first thing that this work shows (in Figure 2 of the paper) is that these real networks are not predicted to be all that stable. When mutualistic interactions are removed, these communities are actually more stable. And when compared to a randomized version of themselves, most real mutualistic networks are also less stable. This second finding is particularly damning, because it really disassembles the idea that these networks exist because they have an architecture that makes them more likely to persist: the extant networks have less stable properties than mixed-up versions of themselves.

The second component of this project looked at the real predictors of stability (in Figure 3 of the paper), starting at the individual level and then moving up to the entire network. The individual survival of a particular species is dictated by the number of mutualistic partners it has. However, the persistence of a mutualistic network is actually more closely related to its “connectance”, a relative measure of how many of all possible mutualistic pairings are actually found in the community network. In addition, “network magnitude” — a measure of how many total possible connections there are in the network — is a particularly good predictor of the persistence in mutualistic networks, as larger networks tend to be less stable.

So what does this all mean? Here is what the authors of this paper have to say:

Together, these results show that any apparent relationship between nestedness and community persistence is a consequence of the correlation of nestedness with simpler properties, such as network magnitude and connectance.

In other words, inasmuch as nested mutualistic networks demonstrate high connectance and low network magnitude, they are persistent. But persistence is — according to these authors — an artifact of other properties of the network, and not their nestedness.

This is something that I am going to need to digest, and I will be interested to read the response of other researchers in this area who have claimed that there is a direct nestedness-persistence connection. The fact that this paper analyzes the properties of real networks instead of hypothesized networks is to its favor, but without fully understanding their methods it is hard to have full faith in their results. If their results withstand scrutiny, they suggest that we need not appeal to overall network stability to understand the ecological communities found in nature; simply understanding the challenges faced by individual mutualist species, whose interest is to maximize their number of partner species, is enough to understand the observed configuration of mutualistic networks.

This paper is unfortunately typical of those published in the “high impact” journals in that it contains most of its explanation in the Supplementary Information. It is a frustrating paper to read because it seems to have one mission (to debunk the idea that nestedness causes ecological networks to be more persistent) and therefore fails to provide much in the way of context or explanation. This paper will get the attention of the small number of researchers doing work in this area, but it will probably be a head-scratcher for most others; if you do not explain why your paper is important, most people will not consider it important.