gustavdelius / mizerevolution

This project adds the possibility to simulate evolutionary processes with the Mizer model

Introduction to mizerEvolution

Overview

The mizerEvolution package is an extension of the mizer package (ref) and simulate evolutionary processes within a size-structured model. Below are described the X steps to use this package wich are described more in details further down.

1. Installing mizerEvolution.

2. Setting the model parameters.

3. Running a simulation.

4. Exploring the results.

Installing mizerEvolution

mizerEvolution is an R package available on GitHub so you first need to have R installed on your computer, need devtools package and then start an R session and type:

devtools::install_github("baldrech/mizerEvolution")

After installing mizerEvolution, you need to load it via library().

library(mizerEvolution)

mizerEvolution is compatible with R versions XX and later. The source code for mizerEvolution is hosted on Github.

Setting the model parameters

As with Mizer, you first need to creat an object of class ? MizerParams. At the moment, the wrapper function evoParams() helps you create this object. It is a tweaked version of newTraitParams which add the necessary parameters to support species evolution, therefore all the default parameters from Mizer are found back in this function. evoParams() adds the lineage parameter which is used to track the ancetry tree of species and the RDD parameter which is used to set the recruitment function. It’s default is extinctionRDD() which allow species to be removed from the ecosystem when they reach an abundance below (10^{-30}ind.m^{-3}).

params <- evoParams(no_sp = 5 )

Running a simulation

This is done by calling the evoProject() function (as in “project forward in time”) with the model parameters. Similar to Mizer’s project(), evoProject() takes the mutation parameter which is the number of new species that are going to be introduced during the simulation length. In the future this parameter will also be a rate of new species appearing in the ecosystem.

sim <- evoProject(params = params, mutation = 1)
#> 2 mutants will be added

This produces an object of class MizerSim which contains the results of the simulation.

Exploring the results

After a simulation has been run, the results can be examined using a range of ?plotting_functions, ?summary_functions and ?indicator_functions. The plot() function combines several of these plots into one:

plot(sim)

In this default Mizer plot, added phenotypes are considered as new species.

Package functionalities

• New species are copy of existing ones albeit with a change to one trait.

• Each projections containing a new species is independent of the others. They are saved in a temporary folder before being binded at the end of the simulation

• New plot functions allow to explore the evolutionary results

Algorithm

Instead of projecting for the entire time one mizer class object, the evoProject() will only project for a shorter amount of time, until a new species needs to be added in the ecosystem. When adding a new species (time chosen randomly), the projection stops, a new species is added with addSpecies(). At the moment, to simulate mutations, the new species is a copy of an existing one albeit for a slight difference in one trait: the maturation size. Further update will include more than one trait (any parameter in the model can become a trait) and mixed reproduction as at the moment new species included this way reproduce independtly from their “parent” and therefore make 2 diffent species. Afte adding a new species, a new projection starts. It means that one simulation is in reality a string of small projection. The finalTouch function takes all theses projection and makes them into one MizerObject wich is usable by any vanilla Mizer functions.

Plotting evolutionary results

base Mizer functions cannot handle species being formed of different “sub-species” so all plot functions have been updated to show species or their “phenotypes” using the lineage parameter

Checking the biomass through time. The default plot has a colorblind of 10 colors and won’t work with more than 10 species, if you want to personalise the plot, use returnData = TRUE.

The following plots are not avaialble on this version yet but are examples of what you can get with biomass data and trait evolution through time.

The biomass per trait value of one species through time

The trait value through time. It’s shown at the species level which is the biomass weighted average of the species’ phenotypes

Looking at the fitness of one species (total spawn output throughtout lifetime) per trait value

When two traits are involved

Checking the average number of phenotypes per species through time.

Add physio plots here

Species invasion

Instead of generating phenotypic diverstity within existing species, the model can instead introduce mutants in an existing ecosystem. One just needs to give a data frame to the mutation argument instead of a numeric.

params<- evoParams()

alien <- params@species_params[4:5,] # copy existing data frame
alien$h <- 50 # change some parameters
alien$alpha <- .6
alien$time <- c(50,75) # when are the invasive species coming in?
alien$lineage <- factor(99,levels = 99) # need to specify a lineage for these species otherwise they will be related to the endemic ones
alien$init_n_multiplier <- NULL # multiplier for the initial abundance

sim <- evoProject(params = params, mutation = alien)
#> 2 mutants will be added
plot(sim)

New plot function have also been introduced to track the change in trait value through time and to calculate the fitness of each species (here being the total spawn output throughout lifetime of a cohort, averaged per individual) to study the effect of the evolutionary processes on the ecoystem

Future updates

• Fast initialisation
• What else?