Last week I somewhat randomly decided that I needed to start assembling a large amount of data. Specifically I was looking for food webs, either an interaction matrix or edgelist preferably in either Excel or .csv file formats (.txt is a bit irritating to put into R, at least with my limited knowledge). The first place I looked was in the ESA Archive Data Papers, and found several relevant sources. In particular I found 5 data papers that included edgelists of predator-prey links, as well as data for various parasitic interaction types.
One of them, a food web from Otago Harbor in New Zealand, I had already been playing around with to get my bearings using network analysis tools in R but the others I downloaded because I like having plenty of data lying around. Ever since I read the 2002 paper by Milo et al. about network motifs, and the subsequent papers by Dan Stouffer on patterns of sub graph representation (here, here, and here) I like to take a look at how motifs are represented in empirical and model networks. In particular I can use R to count the number of times a sub graph occurs in a larger web using the command, graph.count.subisomorphisms.vf2. I used this to get the number of biparallel chains in the food webs I downloaded from the Archives page with and without parasites and found something I think is interesting:
What this shows is that for a given connectance, food webs that include parasites have more biparallel chains than the same webs without. Of course, this is a misleading figure since the parasite webs have more species and significantly more links. By correcting for the number of species it is fairly clear that the difference between the webs is not significant.
Regardless of that misleading adventure I had with the new data it has gotten me very interested in how the inclusion of parasites alters the topological properties of food webs. Importantly, the question arises as to whether there is a significant effect on topology by including parasites, or whether merely sampling predator prey links sufficiently captures the properties of the communities that we are interested in. The inclusion of parasites into food web analyses is a relatively recent endeavor that has taken root in the past five years or so (although there were in fact some researchers investigating parasites and food webs earlier). It is part of the larger ultimate goal of incorporating multiple interaction types into community network analysis to truly capture the dynamics of the system and more importantly to determine the similarities and differences across ecosystem types. A better understanding of community dynamics is going to take us down a path that will help improve conservation and restoration efforts as well as allow us to better model the changes that we should expect as the global climate shifts.
For me, I think that the next step in examining food webs with parasites will be to attempt to quantify the differences between the datasets I have (although comparing networks may prove especially challenging as I will outline in a later post). Then, because these datasets allow me to assign the specific type of parasitic interaction to a link I can investigate how different proportions of specific parasitic interactions lead to topological changes in the web.