Ecological Archives E084-056-A1

Pedro R. Peres-Neto, Donald A. Jackson, and Keith M. Somers. 2003. Giving meaningful interpretation to ordination axes: assessing loading significance in principal component analysis. Ecology 84:2347–2363.

Appendix A. Values of the 95% quantiles for absolute loadings of the first variable in the first five principal components for each correlation structure considered in this study (Fig. 1) and a spherical correlation matrix.

Note that values across components would change according to the chosen variable. Samples were based on normally distributed samples containing N observations. One-thousand samples were used in each case. Parametric correlation coefficients r ( = 0.05) are shown according to sample size. First, note that in all cases, values change along axes. Second, values in the spherical population are always larger than the correspondent tabled correlation value, indicating that the latter may be too liberal. Our results (Tables 2 and 3) show that this is in fact the case. Thus, standard tabled values are in fact inappropriate for asssesing eigenvector significance.



N = 30 (r = 0.361)

N = 40 (r = 0.312)


Component

Component


1
2
3
4
5

1
2
3
4
5
Correlation matrix

Nine variables
Spherical
0.763
0.730
0.709
0.644
0.600

0.735
0.717
0.685
0.665
0.635
1

0.936
0.493
0.323
0.397
0.342

0.937
0.493
0.292
0.391
0.331
2

0.903
0.588
0.372
0.396
0.332

0.889
0.570
0.351
0.403
0.339
3

0.911
0.533
0.361
0.385
0.343

0.900
0.520
0.341
0.380
0.334
4

0.867
0.660
0.480
0.574
0.513

0.846
0.593
0.422
0.580
0.516
5

0.839
0.655
0.515
0.539
0.513

0.822
0.605
0.482
0.568
0.503
6

0.810
0.706
0.626
0.610
0.563

0.799
0.691
0.572
0.613
0.580
7

0.941
0.612
0.383
0.238
0.226

0.936
0.516
0.358
0.196
0.193
8

0.949
0.361
0.250
0.232
0.258

0.946
0.325
0.223
0.195
0.222
9

0.945
0.399
0.251
0.228
0.262

0.944
0.362
0.230
0.195
0.205
10

0.939
0.572
0.275
0.247
0.240

0.941
0.504
0.245
0.209
0.210
11

0.874
0.708
0.478
0.427
0.449

0.864
0.680
0.429
0.406
0.420
12

0.952
0.306
0.267
0.244
0.224

0.949
0.251
0.234
0.206
0.206
13

0.883
0.546
0.446
0.458
0.419

0.875
0.449
0.401
0.382
0.373
14

0.822
0.726
0.591
0.582
0.507

0.818
0.663
0.555
0.549
0.514



N = 60 (r = 0.254)

N = 80 (r = 0.220)


Eighteen variables
Spherical

0.603
0.587
0.573
0.575
0.538

0.581
0.548
0.552
0.532
0.506
1

0.925
0.410
0.241
0.303
0.293

0.924
0.384
0.213
0.297
0.266
2

0.866
0.533
0.307
0.317
0.277

0.856
0.513
0.278
0.279
0.272
3

0.886
0.485
0.290
0.299
0.279

0.876
0.460
0.276
0.296
0.271
4

0.817
0.471
0.297
0.480
0.422

0.814
0.435
0.278
0.442
0.407
5

0.770
0.523
0.341
0.476
0.408

0.761
0.510
0.317
0.435
0.435
6

0.729
0.529
0.395
0.536
0.485

0.728
0.502
0.346
0.496
0.464
7

0.931
0.380
0.233
0.146
0.148

0.927
0.340
0.211
0.134
0.127
8

0.935
0.236
0.150
0.145
0.156

0.933
0.216
0.129
0.122
0.134
9

0.932
0.313
0.158
0.151
0.156

0.927
0.292
0.148
0.122
0.132
10

0.933
0.366
0.160
0.141
0.142

0.931
0.346
0.127
0.130
0.130
11

0.836
0.426
0.315
0.282
0.304

0.824
0.370
0.269
0.252
0.271
12

0.938
0.154
0.138
0.146
0.141

0.937
0.124
0.125
0.129
0.124
13

0.839
0.289
0.281
0.283
0.280

0.829
0.249
0.255
0.253
0.261
14

0.762
0.422
0.398
0.393
0.406

0.750
0.398
0.393
0.394
0.367



[Back to E084-056]