On why and how to incorporate multiple interaction types into network analyses

Submitted by editor on 8 September 2017.Get the paper!

In ecological communities, every species needs to interact with other species in order to survive and prosper. These ecological interactions are extremely varied in typology, magnitude, and mechanisms, as one would guess given the overwhelming diversity of life forms on Earth. Ideally, we could build a mathematical theoretical model to study the dynamics and stability of the whole set of interactions in a community. However, the huge diversity of those interactions constitutes an intricate web or network of interconnections that is extremely difficult to collect and map onto models in a comprehensive fashion. For many study systems, only the most conspicuous interactions are observed, and these usually happen to be competitive, predator-prey, or mutualistic interactions. Hence, most theoretical representations of interaction networks only account for one of these types, assuming implicitly that the role of other interactions is negligible. But that assumption, even if convenient, is at the very best unwarranted, and evidence is mounting that a realistic depiction of interaction networks must account for all the different interaction types altogether.

In our paper, we explore the recent literature on networks considering multiple interaction types and systematize the various methodologies used to represent them. In differentiating these methodological frameworks, we re-encountered a classic debate on the nature of ecological interactions: are interactions to be classified according to their net effect on the species involved (e.g. increase/decrease of species population growth rate), or according to the ecological mechanism by which the interaction takes place (e.g. trophic/nontrophic exchanges)? This dichotomy, alongside the way of quantifying interaction strength, are the two criteria that we propose for differentiating between methodological frameworks. According to these criteria, we describe three frameworks named "expanded food webs", "multilayer networks", and "equal footing networks".

Fig. 1: Three frameworks for analyzing networks with multiple interaction types, exemplified on a small network of 7 species. For details, see the article.

We also review some prominent examples of empirical datasets containing multiple interactions (although difficult, it's possible to sample and assemble empirical networks of multiple types!), and demonstrate the potential applications of each framework with a case study involving some remarkable interactions among plants, their associated pollinators and a species of lizard acting both as a predator on pollinators and plant seed disperser (Fig. 2).

High density of Podarcis lilfordi, a key species in our case study, at the Aire Island (Balearic Islands, Mediterranean Sea). Photograph by Joan Maspons.

With a multiple interactions approach to ecological communities and the increasing availability of empirical observations such as the datasets reviewed in our paper, fundamental questions in community ecology can start to be addressed; for example, how does the ratio between interaction types (and their associated strength) affect community properties, or whether this ratio is preserved across habitats or environmental gradients. Tackling such questions is pressing in order to understand the response of ecological communities to different types and magnitudes of stressors, and we hope that our study will help in systematizing the field and paving the way for addressing them.

The authors through David G Callejas

Twitter: @DavidGCallejas



Insights into Oikos papers