AJ Smit and contributors 2026-02-20

spesim

spesim is an R package for simulating, sampling, and analysing spatially heterogeneous ecological communities in irregular landscapes.
It’s built for teaching, methods testing, and exploratory research in biogeography and community ecology.

Generate individuals for multiple species under realistic species–abundance distributions (e.g., Fisher’s log-series, neutral ZSM).
Impose environmental filtering (per‑species optima/tolerances on named gradients).
Add spatial structure using canonical point processes (Poisson, Thomas, and fast MCMC-based Strauss/Geyer models) or an optional neutral simulation engine with dispersal limitation.
Sample with flexible quadrat schemes (random, systematic, tiled, transect, Voronoi).
Produce site × species matrices, per‑site environment summaries, and diversity partitions (α, β, γ).
Save publication‑quality maps and an optional advanced analysis panel (rank–abundance, occupancy–abundance, species–area, distance–decay, rarefaction).
Write a human‑readable text report that explains what happened.

Why spesim?

Teaching: make abstract ideas (environmental filtering, distance decay, β‑diversity) tangible with quick simulations you can show in class.
Methods development: prototype sampling designs (how many quadrats? which scheme?) and test metrics before going to the field.
Exploration: stress‑test analyses on known “truth” to see how robust your inferences are.

Installation

Development version from GitHub:

# install.packages("remotes")
remotes::install_github("ajsmit/spesim")

No heavy spatstat dependency is required for core functionality. The package relies on common spatial/data/plotting libraries (see DESCRIPTION for imports).

Building the optional pkgdown site locally? Use:
# install.packages("pkgdown")
pkgdown::build_site()

Vignettes (online)

If reading on GitHub before the site is live, you’ll find the sources under vignettes/.

Quick start

You can run a simulation in two ways: programmatically (in-memory) or using an init file on disk. The latter is more convenient for larger setups and reproducibility, since access to the init file gives you a full bird’s eye view of the simulation parameters.

Quick start: programmatic

A minimal run that does not write files (handy for examples/CI).

library(spesim)

# Load a complete example init, then tweak a few things in-memory
P <- load_config(system.file("examples/spesim_init_basic.txt", package = "spesim"))
P$N_SPECIES <- 10
P$N_INDIVIDUALS <- 2000
P$SAMPLING_SCHEME <- "random"
P$N_QUADRATS <- 20
P$QUADRAT_SIZE_OPTION <- "medium"

# Run (skip writing to disk to keep examples snappy)
res <- spesim_run(P, write_outputs = FALSE, seed = P$SEED)

# One-map view: individuals + quadrats (no gradient fill)
p <- plot_spatial_sampling(res$domain, res$species_dist, res$quadrats, res$P)
print(p)

# Optional: a 2×2 panel including gradient overlays
p1 <- plot_spatial_sampling(res$domain, res$species_dist, res$quadrats, res$P)
p2 <- plot_spatial_sampling(res$domain, res$species_dist, res$quadrats, res$P,
                            show_gradient = TRUE, env_gradients = res$env_gradients, gradient_type = "temperature_C")
p3 <- plot_spatial_sampling(res$domain, res$species_dist, res$quadrats, res$P,
                            show_gradient = TRUE, env_gradients = res$env_gradients, gradient_type = "elevation_m")
p4 <- plot_spatial_sampling(res$domain, res$species_dist, res$quadrats, res$P,
                            show_gradient = TRUE, env_gradients = res$env_gradients, gradient_type = "rainfall_mm")
(p1 | p2) / (p3 | p4)

Quick start: init file on disk

Prefer declaring everything in a text file? Point spesim_run() to it:

library(spesim)

# Your own config on disk
init <- "inst/examples/spesim_init_complete.txt"   # or a path you created

# Optional: interactions resolved within the init (via INTERACTIONS_EDGELIST)
#          or via a separate interactions file set here:
# interactions <- "inst/examples/interactions_init.txt"

res <- spesim_run(
  init,
  output_prefix = "out/demo",
  write_outputs = TRUE       # write CSVs, figures, report
)

# Outputs written under out/demo_<timestamp>*
# Results also returned in 'res' for programmatic use:
str(res$abund_matrix)

Typical workflow

Set parameters: either with an init file (see vignette) or in-memory (load_config() → edit fields).
Run the simulator: spesim_run().
Use the outputs:
- res$abund_matrix (site × species),
- res$site_coords (quadrat centroids),
- res$env_gradients (grid with temp/elev/rain fields),
- and map(s) via plot_spatial_sampling().
(Optional) Advanced panel: enable P$ADVANCED_ANALYSIS <- TRUE when you want the multi‑plot diagnostic image written to disk.

Function overview

Function	Purpose
`spesim_run()`	End‑to‑end orchestrator (recommended); returns a `spesim_result` and (optionally) writes CSVs/figures/report.
`run_spatial_simulation()`	Legacy wrapper around `spesim_run()` (deprecated; kept for backwards compatibility).
`generate_heterogeneous_distribution()`	Place individuals using abundances, gradients, dominant clustering, and interactions.
`create_abundance_matrix()`	Build site × species table by intersecting individuals with quadrats.
`calculate_quadrat_environment()`	Mean environmental conditions per quadrat from the grid.
`plot_spatial_sampling()`	Publication‑ready map of domain, individuals, quadrats, optional gradient fill.
`generate_full_report()`	Text report covering gradients, distributions, diversity, and Fisher log‑series validation.
`generate_advanced_panel()`	Patchwork panel: rank–abundance, occupancy–abundance, species–area, distance–decay, rarefaction.

Interactions

You can specify interspecific effects inline in the init file or via a simple CSV edgelist (neighbour effects). See the dedicated vignette:

https://ajsmit.github.io/spesim/articles/spesim-interactions.html

Key points:

Rules are directed (A -> B may differ from B -> A).
Default is neutral (1.0); only list deviations.
Set the global INTERACTION_RADIUS (0 disables the modifier).

Environmental gradients

Three synthetic gradients are available out-of-the-box (temperature, elevation, rainfall), each normalised to [0,1] and reported in common units for plots/reports. You can control per-species optima and tolerances as scalars, vectors (by species), or named by gradient. See the init-file vignette:

https://ajsmit.github.io/spesim/articles/spesim-init-parameters.html

For a quick look at the generated grid:

domain <- create_sampling_domain()
env <- create_environmental_gradients(domain, resolution = 50, noise_level = 0.05)
head(env)

Citation

If you use spesim, please cite:

@misc{spesim,
  author = {AJ Smit},
  title  = {spesim: Spatial Ecological Simulation in R},
  year   = {2025},
  url    = {https://github.com/ajsmit/spesim}
}

License

MIT License — see LICENSE.