Scalable Joint Species Distribution Modeling
Install sjSDM and its dependencies
Installation help
getCor
getCov
getImportance
Post hoc calculation of standard errors
Get weights
Importance of environmental, spatial and association components
install diagnostic
AccSGD
AdaBound
Adamax
Anova / Variation partitioning
biotic structure
check module
check model check model and rebuild if necessary
Return coefficients from a fitted sjSDM model
DiffGrad
Non-linear model (deep neural network) of environmental responses
Generate spatial eigenvectors
Print a fitted sjSDM model
Plot internal metacommunity structure
is_torch_available
Linear model of environmental response
Extract negative-log-Likelihood from a fitted sjSDM model
madgrad
new_image function
Plot elastic net tuning
Training history
Coefficients plot
Plot anova results
Plot importance
Plot internal structure
Plot predictors of assembly processes
Internal coefficients plot
Predict from a fitted sjSDM model
Print a bioticStruct object
Print a DNN object
Print a linear object
Print a fitted sjSDM_cv model
Print sjSDM anova object
Print importance
Print internal structure object
Residuals for a sjSDM model
RMSprop
R-squared
Set weights
SGD
Simulate joint Species Distribution Models
Generates simulations from sjSDM model
Cross validation of elastic net tuning
Fitting scalable joint Species Distribution Models (sjSDM)
sjSDM control object
Return summary of a fitted sjSDM_cv model
Return summary of a fitted sjSDM model
Summary table of sjSDM anova
Update and re-fit a model call
A scalable and fast method for estimating joint Species Distribution Models (jSDMs) for big community data, including eDNA data. The package estimates a full (i.e. non-latent) jSDM with different response distributions (including the traditional multivariate probit model). The package allows to perform variation partitioning (VP) / ANOVA on the fitted models to separate the contribution of environmental, spatial, and biotic associations. In addition, the total R-squared can be further partitioned per species and site to reveal the internal metacommunity structure, see Leibold et al., <doi:10.1111/oik.08618>. The internal structure can then be regressed against environmental and spatial distinctiveness, richness, and traits to analyze metacommunity assembly processes. The package includes support for accounting for spatial autocorrelation and the option to fit responses using deep neural networks instead of a standard linear predictor. As described in Pichler & Hartig (2021) <doi:10.1111/2041-210X.13687>, scalability is achieved by using a Monte Carlo approximation of the joint likelihood implemented via 'PyTorch' and 'reticulate', which can be run on CPUs or GPUs.
Useful links