Skip to contents

Predicted Effects of Environmental Gradients on Dissimilarity

diss_gradient.Rd

Computes the predicted effect of one or more covariates on compositional dissimilarity from a fitted gdmm or bbgdmm model. Optionally, returns uncertainty bands (credible or confidence intervals) using either bootstrapped samples or parametric bootstrapping.

Usage

diss_gradient(
  m,
  var = "all",
  n = 100,
  CI = TRUE,
  n_sim = NULL,
  CI_quant = c(0.95)
)

Arguments

m

A fitted model object of class gdmm or bbgdmm.

var

Character vector of variable names to evaluate. Use "all" (default) to use all predictors in the model.

n

Integer. Number of points to evaluate along each gradient. Defaults to 100.

CI

Logical. Whether to compute confidence intervals. Default is TRUE.

n_sim

Number of draws used to obtain CI. If NULL, uses n_boot for bbgdmm objects or 1000 for gdmm.

CI_quant

Numeric vector specifying the quantile coverage (e.g., 0.95 for 95% CI). Default is 0.95.

Value

A named list of data frames, one for each variable in var. Each data frame contains:

  • var: Variable names.

  • f_x: Predicted value after fitted non-linear transformation (e.g., monotonic I-splines).

  • x: The values of the focal gradient.

  • CI lower / CI upper: (Optional) Lower and upper bounds of the confidence interval.

Details

The function generates predicted dissimilarities along a gradient of each specified predictor, while holding all other predictors constant. If CI = TRUE, it uses bootstrap samples (for bbgdmm) or simulates from the joint covariance matrix (for gdmm) to construct confidence or credible intervals.

This approach is useful for visualizing directional compositional changes along individual environmental gradients.