bjNMDS: Multiple NMDS Ordinations from Beta Distributed Jaccard...

In jarioksa/natto: An Extreme 'vegan' Package of Experimental Code

Multiple NMDS Ordinations from Beta Distributed Jaccard Dissimilarity

Description

Function performs metaMDS on expected Beta Jaccard dissimilarity with multiple starts and then a number of monoMDS runs on random Beta Jaccard dissimilarities starting from the expected configuration, and Procrustes rotates the random solutions to the expected one.

Usage

bjNMDS(
  x,
  n = 100,
  trymax = 500,
  maxit = 1000,
  smin = 1e-04,
  sfgrmin = 1e-07,
  sratmax = 0.999999,
  parallel = 2,
  trace = FALSE,
  ...
)

## S3 method for class 'bjnmds'
plot(
  x,
  choices = 1:2,
  kind = c("hull", "ellipse", "wedge", "star"),
  keep = 0.9,
  type = "t",
  ...
)

Arguments

x	Community data to be analysed and treated as binary (bjNMDS) or object to be plotted (plot).
n	Number of random samples of Beta Distribution.
trymax	Maximum number of random starts in metaMDS.
maxit, smin, sfgrmin, sratmax	Convergence parameters in monoMDS.
parallel	Number of parallel tries in metaMDS.
trace	Trace iterations in metaMDS.
...	Other parameters passed to functions.
choices	Axes to be plotted.
kind	Shape to be plotted to show the scatter of coordinates of sampled distances; see bjpolygon for details.
keep	Proportion of points to be enclosed by shape; see peelhull and peelellipse.
type	Type of the plot: "t"ext, "p"oints or "n"one.

Details

Current function is designed for visual inspection of random variation of ordination, and no numerical analysis functions are (yet) available. The plot can show the scatter of points as convex hulls or ellipsoid hulls containing a given proportion of random points, or as stars that connect the expected point to randomized ones. With option "wedge" the convex hull is extended to include the observed point if necessary. See bjpolygon and bjstars for technical details.

Missing functionality includes adding species scores, adding fitted environmental vectors and factors and numeric summaries of the randomization results.

Value

A metaMDS result object amended with item rscores that is a three-dimensional array of coordinates from random bayesjaccard ordinations.

Examples

data(spurn)
m <- bjNMDS(spurn)
plot(m, keep = 0.607, col = "skyblue")

jarioksa/natto documentation built on March 28, 2024, 12:45 a.m.