A model of trait-based/neutral dynamics of trophic interaction networks

Author: Timothée Poisot (t.poisot_at_gmail.com)

Compilation and execution requires the GNU Scientific Library.

Released under the terms of the GPL license. Forking this repository grants implicit authorisation to pull and merge all future changes and release them (with attribution) under the original license.

The default usage is to compile the program from source using make. This will put to executables, niche and pop, in the bin/ folder, and create an output/ folder in which pop will write. Both niche and pop should be launched from within the bin/ folder.

niche is a implementation of the niche model of food webs, also generating random carrying capacities based on species body sizes (see the preprint discussing this model to see how this is done).

niche S C

will generate a network of (exactly) S species, with (approximately) C connectance. The resulting list of species will be put in a file called splist.txt. This file is then read by the sub-program pop.

The file splist.txt is made of 6 columns: the identifier of the species (from 1 to S), its body size, range, and centroid (all between 0 and 1, as in the original niche model of Williams & Martinez), and the randomly generated carrying capacity and starting population size.

pop is an individual-based simulation model using the output of niche to simulate network dynamics. Note that because pop reads informations within a file called splist.txt, the user can generate this file in any other way. Examples are given in the publication accompanying this program.

The executable generated after compilation requires a number of arguments:

1. the file in which the species list is stored (splist.txt, unless another file was generated)
2. an integer giving the number of species in the flile splist.txt
3. an integer telling whether to use body size to determine feeding range (1) or only neutral dynamics (0)
4. an integer giving the simulation time
5. an integer giving the number of timesteps that should be recovered by the end of the simulation
6. the prefix of the output file
7. an integer telling whether there is migration at each timestep (1for migration, 0 for no migration)
8. an integer telling how many individuals are added at each time step
9. an integer telling whether the system starts with all species (0), or whether it should be assembled species-by-species (1)

The pop sub-program generates json files, following the following scheme:

{
   "name":"argument 6",
   "species":{
      "0":{"n": 0.84, "r": 0.02, "c": 0.61, "K": 183},
      "1":{"n": 0.11, "r": 0.10, "c": 0.08, "K": 504},
      ...
   },
   "times":[
      "0":{
         "pop":{
            "0": 121,
            "1": 104,
            ...
         },
         "int":[
         {"pred": 0, "prey": 4},
         {"pred": 3, "prey": 3},
         ...
         ]
      }
   ]
}

The species arry has S elements, containing the species identifier, the niche model parameters n, c and r, and the carrying capacity K. The times array has one element per recorded timestep, with each entry being composed of an array pop (identifier of the species and current population size), and an array int (identifier of the predator and the prey). Take together, these elements are enough to reconstruct (i) population sizes and (ii) network structure over time. This object can then easily be parsed using JSON parsers in various languages (json in Python, rjson in R).