Description Usage Arguments Details Value Note Author(s) References See Also Examples

Function for determine the number of significant dimensions in principal coordinate analysis (PCoA).

1 2 3 4 5 6 7 8 9 10 11 12 | ```
pcoa.sig(data, method = "gower", squareroot = FALSE, axis = 6,
n.start = NULL, by = 1, iterations = 1000, parallel = NULL,
newClusters = TRUE, CL = NULL)
## S3 method for class 'pcoasig'
print(x, ...)
## S3 method for class 'summarypcoasig'
print(x, ...)
## S3 method for class 'pcoasig'
summary(object, choices = c(1, 2), ...)
``` |

`data` |
Community data matrix. |

`method` |
Method for dissimilarity index, as accepted by |

`squareroot` |
Logical argument (TRUE or FALSE) to specify if use square root of dissimilarity index (Default squareroot = FALSE). |

`axis` |
Maximum number of ordination principal axes to be monitored (Default axis = 6). |

`n.start` |
Initial sample size. If n.start = NULL initial sample size is equal to total sample size (Default n.start = NULL). |

`by` |
Sampling unit is added at each sampling step (Default by = 1). |

`iterations` |
Number of permutations to assess significance (Default iterations = 1000). |

`parallel` |
Number of parallel processes. Tip: use detectCores() (Default parallel = NULL). |

`newClusters` |
Logical argument (TRUE or FALSE) to specify if make new parallel processes or use predefined socket cluster. Only if parallel is different of NULL (Default newClusters = TRUE). |

`CL` |
A predefined socket cluster done with parallel package. |

`x` |
An object of class pcoasig. |

`...` |
Other parameters for the respective functions. |

`object` |
An object of class pcoasig. |

`choices` |
Axes for re-scaling. Choices must have length equal to two (Default choices = c(1, 2)). |

At each iteration step a bootstrap sample is subjected to PCoA ordination, the scores are submitted to a procrustean adjustment, and the correlation between observed and bootstrap ordination scores is computed. It compares such correlations to the same parameter generated in a parallel bootstrapped ordination of randomly permuted data. The number of axes in bootstrap or null PCoA with eigenvectors corresponding to positive eigenvalues may be smaller than the number of axes monitored, in this case, axes with values equal to 0 are created. The number of iterations with original values for each axis is shown in n.permut.bootstrap and n.permut.null.

The function scores.pcoasig re-scales the correlation values for `biplot`

graphics.

`value` |
The eigenvalues, relative eigenvalues and cumulative relative eigenvalues.. |

`vectors` |
The principal coordinates. |

`correlations` |
Correlations between axis and original data. |

`mean.cor.null` |
Mean correlations, for axis, between null and reference scores. |

`mean.cor.bootstrap` |
Mean correlations, for axis, between bootstrap and reference scores. |

`n.permut.bootstrap` |
Number of iterations for each axis in bootstrap step. |

`n.permut.null` |
Number of iterations for each axis in null step. |

`probabilities` |
Probabilities for each axis. |

**Principal Component Analysis (PCA)**

You can use the same function to determine the number of significant dimensions in principal component analysis (PCA). For this, standardize each variable for zero mean and uni variance (function decostand and method standardize) and use euclidean distance as dissimilarity index.

**Interpretation**

If the higher dimension is significant, then all lower dimensions will also be significant.

Vanderlei Julio Debastiani <[email protected]>

Pillar, V.D. (1999). The bootstrapped ordination reexamined. Journal of Vegetation Science 10, 895-902.

1 2 3 4 |

PCPS documentation built on May 24, 2018, 5:04 p.m.

Embedding an R snippet on your website

Add the following code to your website.

For more information on customizing the embed code, read Embedding Snippets.