FoodChain

Building food webs - simple or complex?
In my last post, I explained why resolution matters in food webs. However, I never properly introduced what is a food web and how to build them.
A food web is a graphic representation of predator-prey relationships, in other words 'who eats whom'. It is also a generalization of the food chain concept. Not only do we represent the flow of energy going from one primary producer to a top predator, but also every single food chain in the community. At least in theory, we try to build the food web as complete as possible. In practise, monitoring every species and their interactions in an ecosystem is challenging if not impossible. Food webs have graphically a long history. The first such representation reported in the literature dates from 1880. Camerano (1880) represented the relationships between beetles; what he called "their enemies" (i.e. their predators) and the enemies of those enemies (Fig. 1). The representation was rather simple: lines connecting a beetle to a predator who would be connected to a second predator and so on; each line representing a food chain this beetle was involved in.

At that time, a food web was a sole representation of relationships between species. We had to wait almost 50 years until the work of Charles Elton (1927), for food webs to become a more practical tool. He attempted to represent each species and each of their relationships in what he called "food cycles". Almost 100 years later, ways of analysing food webs may have changed but the old diagram still remains...with its share of issues: "How to draw complete food webs?" and if not possible, "how to make them model what should be their exhaustive counterparts?" For representing complete food webs, one must identify every single species in the community (i.e. the species composition) together with 'who eats whom' (i.e. their trophic links). Although the task may seem simple, the more species, the more possible interactions.
Let's do some simple maths. First, we will consider the following conditions:
(1) species-to-species predation (i.e. species A feeding on species B, B feeding on A; and A and B feeding on themselves);
(2) no mutual predation (i.e. meaning we exclude B feeding on A from the first condition)
(3) no cannibalism (i.e. excluding species feeding on themselves, A feeding on A)

Under those conditions, for 10 species only, there exists 45 possible interactions. If we forget the above exclusions, there would exist 100 possible interactions. Now let's be more rational: there exist a lot more than 10 species in an ecosystem. For example, the Barents Sea food web I previously presented, contained about 233 trophospecies (Olivier and Planque 2017). I let you sit down in a nice armchair and do the maths. Yes. Exactly. That's a lot of possible interactions!
NOTA BENE:
To compute the number of interactions, first consider there exists at most S2 possible interactions (e.g. for two species A and B there exists 4 possible interactions: A feeds on B, B feeds on A, A feeds on A; and B feeds on B). S represents, the number of species. If we exclude cannibalism, we exclude S interactions. If we don't consider mutual predation, then only half of the interactions are considered (i.e. A feeds on B and we exclude B feeds on A). We are left with the following equation: (S2 - S)/2. Rather simple.

Trophic links can be collected in two main ways: either you observe those interactions yourself, or you find someone who did. In other words, (1) we can collect species trophic interactions from feeding ecology studies (e.g. stomach content analysis shown on Fig 2., food preference experiments); or (2) from the literature based on here-above knowledge in feeding ecology. I personally did both. The former requires strong expertise on organisms found in the community. As a consequence, we usually focus on one particular species (e.g. Clupea harengus), or a group of species (e.g. fish) but we are usually not experts on every species in the community.
Stomach content analysis
Figure 2. Animation showing the stomach content analysis of European perch (Perca fluviatilis)
As a consequence, food web scientists rely on the expertise of their peers. A large part of building a food web lies in doing an extensive literature review to identify realized and potential links. Sometimes, links are missing. The information has not been collected yet and may require inferring species diet; or worse, to lump species together whether they have exactly the same diet or not (Jordán 2003). Yet incomplete, food webs remain essential: if properly built, they give the first insight in functioning of the community. They can help for example (1) identifying keystone species or (2) follow toxic chemicals and microplastics from one species to another. Simple? Complex? Building food webs relies on the multiplication of many rather simple tasks. The more complete, the more challenging. Nevertheless, the result is always rewarding.

References
Camerano, L. 1880. Dell’equilibrio dei viventi mercé la reciproca distruzione. - Accademia delle Scienze di Torino 15: 393–414.
Elton, C. 1927. Animal ecology. - Sidwick and Jackson.
Jordán, F. 2003. Comparability: the key to the applicability of food web research. - Appl. Ecol. Env. Res. 1: 1–18.
Olivier, P. and Planque, B. 2017. Complexity and structural properties of food webs in the Barents Sea. - Oikos 126: 1339–1346