In matthewkling/stranger: Demographic species range modeling with dispersal

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "man/figures/README-",
  out.width = "100%"
)

DERANGED: DEmographic RANGE modeling with Dispersal

This R package provides tools for simulating species geographic range dynamics. Range simulations combine a demographic module that simulates local population change as a function of species characteristics and environmental conditions, and a dispersal module that simulates movement among grid cells.

This is a project of Barracuda, an NSF-funded research program focused on modeling landscape adaptation to climate change. The package is still under development, and API changes and additional features will be implemented over time.

Installation

You can install the development version of deranged from GitHub with:

# install.packages("devtools")
devtools::install_github("matthewkling/deranged")

If you're on MacOS and encounter compilation errors when attempting to install the package, installing the latest version of gfortran may solve the issue.

Example

In this toy example we'll run a range simulation for an imaginary plant species. We'll use a stage-based population matrix model representing three life history stages: seeds, juveniles, and adults.

Input data

First we'll mock up some spatial data. The convenience function landscape_template generates a skeleton object with the necessary data structure, a list with two components. The first piece, n, is a 3-D array of initial population sizes for each life history stage for each spatial grid cell; here we'll start with 10 adults in one of the grid cells. The second piece, e, is a 4-D array of environmental data, which maximally can include multiple environmental variables that vary spatially and temporally; in this example we'll use a single, time-invariant environmental grid with values ranging in a spatial gradient from 0 to 1.

library(deranged)
set.seed(123)

stages <- c("s", "j", "a") # names for life history stages
ls <- landscape_template(n_row = 25, n_col = 25, names = stages) # spatial data template
ls$n[10,10,3] <- 10 # start with 10 adults in one grid cell
ls$e[] <- seq(0, 1, length.out = prod(dim(ls$e))) # set environmental values

Species parameters

The other input to the simulation is a set of parameters representing the species' biology. This includes demographic parameters and dispersal parameters. Here we'll use species_template to create a parameter object, and then set some of the parameter values to fit our species.

For this matrix model, the demographic parameters represent probabilities of transitioning from one stage to another at each time step; these probabilities can be constant, or they can vary as a function of population levels (density dependence) or environmental values (environmental dependence). A given transition probability is calculated as the intercept term from the alpha matrix, plus the coefficients from the beta array multiplied by population levels, plus the coefficients from the gamma array multiplied by environmental values. In this example we'll make juvenile survival depend on the environment, and we'll make juvenile-adult transitions depend on the number of adults.

The other parameters control dispersal distances. Users can supply any arbitrary probability density function, but for now we'll stick with the lognormal distribution included in the template and just modify its parameter values.

sp <- species_template(n_env = dim(ls$e)[4], names = stages)

# invariant transitions
sp$alpha["j", "s"] <- .01
sp$alpha["a", "a"] <- .9
sp$fecundity["a"] <- 100

# environmentally dependent juvenile survival
sp$alpha["j", "j"] <- .1
sp$gamma["j", "j", "v1"] <- .8

# density dependent juvenile-adult transition
sp$alpha["a", "j"] <- .1
sp$beta["a", "j", "a"] <- -.001

# parameters for lognormal dispersal kernel
sp$kernel$params$L <- .3
sp$kernel$params$S <- 3

Simulation

Using these landscape and species inputs, we can now simulate range dynamics over time. At each time step, the species goes through a demographic transition, and then dispersal occurs. The function returns an array of populations over space and time.

d <- simulate(sp, ls, # data and parameter inputs from above
              n_steps = 200, # number of iterations to simulate 
              record = 3) # record populations for life stage 3 (adults)

We can quickly visualize the output using a couple plotting functions. The first plot below shows population time series for 10 random grid cells, with populations in many cells leveling off due to density dependence. The second plot shows snapshots of spatial population patterns over time, with an east-west gradient in population size resulting from the spatial environmental gradient.

library(tidyverse)
d %>% plot_lines() %>% print()
d %>% plot_heatmaps() %>% print()

matthewkling/stranger documentation built on Feb. 25, 2024, 2:31 p.m.