Hint The argument parallel = TRUE is used to greatly accelerate the
computation with large data sets. It calls upon package parallel.
Clusters enclosed in red boxes and underlined correspond to significant AU pvalues (p ≥ 0.95), whereas clusters enclosed in blue rectangles show less
significant values (p ≥ 0.91 in this example). This analysis is useful to highlight
the most “robust” groups of sites. Note, however, that for the same data the
results may vary from run to run because this is a procedure based on random
resampling
.
Once we have chosen a dendrogram and assessed the robustness of its clusters, let
us examine three methods that can help us identify an appropriate number of groups:
silhouette widths, matrix comparison and diagnostic species.
4.7.3.4 Average Silhouette Widths
The silhouette width is a measure of the degree of membership of an object to its
cluster, based on the average dissimilarity between this object and all objects of the
20
21
22
26
27
28
29
30
25
23
24
5
9
15
16
19
17
18
1
13
14
11
12
2
3
7
10
4
6
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
Cluster dendrogram with AU/BP values (%)
Distance: euclidean
Cluster method: ward.D2
Height
97
86
72
99
78
89
99
99
67
84
62
94
60
82
88
88
97
92
97
74
91
83
69
91
95
88
81
au
80
39
43
62
49
64
99
81
37
44
35
73
19
47
77
65
44
59
76
26
27
15
18
9
14
34
25
bp
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
edge #
Fig. 4.10 Multiscale bootstrap resampling applied to the Chord-Ward dendrogram
80
4 Cluster Analysis