Measures on output
These functions let you work with the output of the simulations (i.e.
the object returned by simulate).
Population variability
In the original paper {{ "brose" | cite }}, population variability is defined
as the average of the negative coefficients of variations of biomasses of
persisting species. This can be measured using the population_stability
function. There are two arguments you can act on: last (the number of
timesteps before the end on which the stability should be measured), and
threshold (the biomass under which populations are considered extinct for
the purpose of this measure).
This measure is extremely sensitive to both of these parameters, especially when the system does not reach a stable equilibrium. We found that a way to give much more stable results is to consider all species, even those with very small biomasses. This can be done by setting a negative threshold. Usually, at least 1000 timesteps are required to get a stable estimate of stability.
Note that in the original paper, what is presented is actually this measure, multiplied by 100, which is the relative standard error, and not the coefficient of variation. Note also that, so as to correct for the fact that the number of timesteps varies, we use the corrected estimator of the coefficient of variation.
Population biomass
The population_biomass function returns the average biomass over last
timesteps for every population in the network.
Total biomass
The total_biomass function returns the total biomass over last
timesteps for the entire network.
Food web diversity
The foodweb_diversity is the Shannon entropy measure, corrected for the
number of population (i.e., divided by the natural log of the number of
populations). Values of 1 indicate high evenness, and values close to 0
indicate extreme un-evenness. In the original paper, diversity is measured
as the number of species with a biomass above a given threshold. Given that
this threshold has to be set in an arbitrary way, and does not account for
the fact that changing several parameters also changes the distribution of
biomasses, we have not retained this measurement of diversity.
Saving the simulations
The object returned by simulate can be saved using the befwm.save
function. This function is not exported, so it must be called with the
befwm. prefix. By default, this function will generate a unique identifier
for every simulation. It is generally considered good practice to save the
simulation outputs, and process them later, rather than working on the objects
without saving them. In case one wants to return to the simulations at a
later time, saving the objects allows to forgo re-running the simulations,
and can therefore save significant amounts of time.