A new article was published this past month in PLoS ONE by Elsa Canard and colleagues about how neutral dynamics can shape the structure of trophic networks. They note that simply by assuming that species are interacting at random, and therefore the probability of any two species interacting is a function of the abundances of each species. Thus a given individual is more likely to interact with a species that has a large abundance as compared to a rare species. Low species abundances can therefore generate what Canard et al term “neutral forbidden links.”
I find this type of thinking to be really interesting because it goes against the natural inclination of most ecologists, which is to assume that there are some underlying heuristic rules ordering the structure of networks like food webs (e.g. niche, cascade, nested-hierarchy models). While the authors of this study were able to generate food webs that had realistic structural patterns (compared to a number of empirical networks as well as a null model) they make it clear that they do not advocate the idea that neutral processes are the only forces shaping food webs. Nonetheless they make a good argument for including neutral dynamics when considering network dynamics.
In most of the papers I have read that use models to investigate processes driving network topology and dynamics, there is always some comparison of the model derived network to a null model (usually that of a random or psuedorandom network). I think that it would be useful to incorporate neutral based models as an additional comparison. By comparing model networks to those derived by neutral processes we can tease apart the effects of the mechanism of interest on the network from those that are unrelated to the mechanism.
Jeremy Fox over at Dynamic Ecology just posted an interesting discussion of null models asking whether or not they really nullify what the researcher intends to nullify. He also mentions that some null models, such as the neutral model of evolution truly do nullify selection (setting s=0). Perhaps the model proposed by Canard and colleagues may provide a valuable “neutral theory of food web development” in ecology.
What do you think?