An R package to simulate species abundances (counts) along gradients
One of the key ways quantitative ecologists attempt to understand the properties and behaviour of the methods they use or dream up is through the use of simulated data. There are a number of computer programmes for simulating ecological data along gradients, such as Peter Minchin's COMPAS, but none (that I am aware of) that are available for R on CRAN. Dave Robert's coenoflex package for R would be a useful alternative but currently is archived on CRAN because of some problems in the Fortran code underlying the package.
Rather than have to reinvent the wheel each time I wanted to simulate some new data for a paper or to work on a new approach, I decided to start my own R package to contain a range of simulators encapsulating different response models, numbers of gradients, etc.
At the moment, coenocliner is limited in what it can do practically. There is a single response model, the Gaussian response, which is a symmetric model of the parameters; the optimum, tolerance and height of the response curve. Count data can be generated from this model from either a Poisson or negative binomial distribution, using the parameterised Gaussian response as the expectation or mean of the distribution.
Additional response models include:
- The generalised beta response function
A further feature of coenocliner that I hope to develop is to
include simulation wrapper functions that replicate the simulation
methods used in research papers. A working example is simJamil
,
which produces simlations from a Gaussian logit response following
the scheme descibed in Jamil & ter Braak (2013).
I would like to see coenocliner be as inclusive as possible; if you have code to simulate ecological species or community data that is just sitting around, consider adding it to coenocliner. In the meantime, I'm happy just having something tangible for my own use without having to remember the expressions for some of the response models.
Currently coenocliner is licensed under the GPL v2, but I'm happy to reconsider this if you want to contribute code under a more permissive licence.
No binary packages are currently available for coenocliner. If you have the correct development tools you can compile the package yourself after downloading the source code from github. Once I work out how to link git with svn I'll start a project on R-forge which will host binary packages of coenocliner.
If you use Hadley Wickham's devtools package then you can install coenocliner directly from github using functions that devtools provides. To do this, install devtools from CRAN via
install.packages("devtools")
then run
require("devtools")
install_github("coenocliner", "gavinsimpson")
Jamil and ter Braak (2013) Generalized linear mixed models can detect unimodal species-environment relationships. PeerJ 1:e95; DOI 10.7717/peerj.95