dimcheckMDS: Stress plot/Scree plot for NMDS

Description

This function provides a simple plot of stress values for a given number of tested dimensions (default k = 6) in NMDS. This stress plot (or scree plot) shows the decrease in ordination stress with an increase in the number of ordination dimensions. It is based on function metaMDS (vegan package) and uses the monoMDS engine.

Usage

dimcheckMDS(matrix, distance = "bray", k = 6, trymax = 20, autotransform = TRUE)

Arguments

matrix

Community data, a matrix-like object with samples in rows and species in columns.

distance

Dissimilarity index used in vegdist.

Number of dimensions.

trymax

Maximum number of random configuration for iterative search search of stable solution.

autotransform

Whether to use transformation (see metaMDS) or not. Default is autotransform = TRUE.

Details

Goodness of Non-metric multidimensional scaling (NMDS) is measured by stress value. The lower the stress value, the better fit of original distances/dissimilarities and projected distances in ordination diagram is reached. Stress value depends on dimensionality; it is decreasing with increasing dimensionality. On the other hand, stress-reduction does not mean to maximise interpretation capability. Low-dimensional projections are often better to interprete. and are so preferable for interpretation issues. The stress plot (or sometimes also called scree plot) is a diagnostic plots to explore both, dimensionality and interpretative value. It provides dimension-dependant stress reduction and curve estimate gives indices for meaningful stress reduction with increasing dimensionality. Furthermore, another diagnostic plot for detecting best dimension for projection of NMDS, the Shepard diagram (stressplot) is recommended for detecting best dimensionality in NMDS. Clarke 1993 suggests the following guidelines for acceptable stress values: <0.05 =="" excellent,="" <0.10="good," <0.20="usable,">0.20 = not acceptable. The plot shows the border of the 0.20 stress value limit. Solutions with higher stress values should be interpreted with caution and those with stress above 0.30 are highly suspect.

References

Clarke, K. R. (1993). Non-parametric multivariate analysis of changes in community structure. Austral J Ecol 18: 117-143.

Examples

Run this code

## Use of function with default values
dimcheckMDS(schedenveg)

## Use of function for testing 10 dimensions
dimcheckMDS(schedenveg, k = 10)