Communities are composed of microbial taxa that necessarily interact among themselves, either in positive or negative ways. Understanding such interactions will eventually help in manipulating communities for desired purposes. For example, in bioremediation, in biotechnology, in human health and even in terraforming. We use three different approaches to identify interacting microorganisms.
One approach consists of building co-occurrence networks. The assumption is that taxa that appear together a number of times significantly above what could be expected by chance are likely partners of an interaction. We use an approach that sequentially builds modules of two to 10 or so taxa that co-occur and then we look for their genomic potential to see whether they can complement each other or compete with each other. We have used the extensive information in our data base EnvDB (link) to build such networks for many different environments, including human gut, animal associated microbiotas, aquatic or terrestrial samples. Many of the environments are very rich comprising thousands of species.
A second approach also uses co-occurrences of microbial taxa. However, in this case, the abundance of the different taxa is followed through time. Microbes that oscillate in abundance in concert, are likely to interact with each other. This information can be extracted with generalized Lotka-Volterra equations. We use this approach in an extreme environment: the crystallizer ponds in solar salterns. The microbial community in these environments is extremely simple, including only 10 to 20 species and, thus, making the analysis possible.