gvdr/daedalus-web

Are Food Webs complex? And what does even mean to be complex?

daedalus-web's Issues

Information theory

I'd like to do some information theory things in this paper -- which networks have more informativity? What is the distribution of species contributions? ...

Notes from our discussion

The current null model approach fixes connectance and link density.
Can we develop a null model based on a sampling species from a regional web. This would allow connectance and link density to vary. Could be based on a well resolved empirical dataset or could be simulated.

Here is the paper the I mentioned that uses process based species pools - http://jeanphilippelessard.com/wp-content/uploads/2015/11/Lessard-et-al.-2016.-Am.-Nat..pdf

Allison notes on plenary chat:

(sorry Allie, I meant to rewrite them, but I'm too lazy and I will just copy/paste them)

On the call: Giulio, Daijiang, Tim, Allie, Leonardo

• to do today: 1) what will be our take home message? 2) who is our audience? 3) how will we write the manuscript?

On complexity

• 'complexity' can be many different things at the same time, representing many different facets

• some facets of complexity easy to compute/formalize, some are proxies of complexity (are correlated with it)

• what are the different meanings of complexity that folks talk about? can classify the different definitions/clarify the semantics and then define the meanings mathematically

• there's an implicit hierarchy of complexity metrics: 1) number of species 2) connectance 3) some metric of the different types of interactions in the community (beyond food web) 4) topological, structural, dynamical metrics

• ecologists see a lot of species and assume the web is complex - is that true? could be that most species are interacting weakly - is that web then complex?

• but, on the other hand, weak/rare interactions not the same as having no influence, wrt information

• unpredictability as a metric of complexity? (unpredictability as not knowing initial conditions/structure of the network), though random processes are also unpredictable.

• assuming we know the "true" network - then how do we quantify complexity: is there a universal scale of complexity? or is it always relative to a "random" or "null model" network?

• are all communities/food webs complex? is it a binary?

• why is complexity relevant to ecologists? not because it is an indicator of stability, but maybe through its relationship to evolution/learning

• could a metric of complexity incorporate the evolutionary past of the network? in other words, are food webs with a shorter history less complex? or are they more complex? for example, a pair of interacting species that just recently diverged might have a less predictable (more complex?) interaction than a pair that diverged long ago. nontrivial to introduce these ideas into a metric of complexity

• exploring the comparison between observed & randomized networks: for some datasets, can’t randomly generate the same properties by chance. does that mean it is complex? (allie note: is there a connection to kolmogorov complexity here?)

• complexity as informational complexity: a complex system is somewhere between complete regularity and complete randomness

Audience

• we all are in some way unhappy with the way ecologists talk about complexity, so the target audience would be community ecologists likely to use network theory

• perhaps write an opinion paper with some numerical experiments to back up/illustrate ideas

• how to reach out to different communities of community ecologists (beyond food web types)