Analysis of some Arctostaphylos samples in Charlton Flat / Chilao / Three Points Area
Keir Morse and Nancy Accola gathered eight samples of twigs from the manzanitas all the authors studied from the Charlton Flat / Chilao / Three Points area, for use in comparing plants in the field. Rather than just tossing those samples at the end of the day, Tom measured them up to see how well a PCA could separate them.
First, Tom lined up the samples in order of leaf color, to estimate how green the leaves were for each sample. The samples are shown in Fig. 1, with each sample numbered. Samples #1 and #3 are labeled as ssp. mollis, from the results of the following analysis. We don't know yet the determination of the other samples.
Fig. 1. Photographs of the samples in order of estimated color of the leaf, from left to right in top row, followed by left to right in bottom row. Click on the picture for a larger version.
Closer pictures of each sample, and a close-up of the stem hairs on the twigs, are linked in Table 1.
The close-up of the stem hairs shows the section of the stem in each sample which has the highest density of long hairs. Thus even though some samples have no long hairs on the majority of their internodes, the link in Table 1 for those samples shows the un-representative internodes with long hairs. This was done because some keys use the presence of any long hairs to separate out ssp. mollis.
The measurements for each sample are given in Table 2.
Table 1. Links to more detailed pictures for each sample
| Sample # | Entire Sample | Twig hairs |
|---|---|---|
| #1 | sample | stem hairs |
| #2 | sample | stem hairs |
| #3 | sample | stem hairs |
| #4 | sample | stem hairs |
| #5 | sample | stem hairs |
| #6 | sample | stem hairs |
| #7 | sample | stem hairs |
| #8 | sample | stem hairs |
Table 2. Measurements on the Samples (note the samples are grouped by long hairs, not # order)
| Characteristic | Many Long Hairs | Some Long Hairs | No Long Hairs | |||||
|---|---|---|---|---|---|---|---|---|
| Sample # | #1 | #3 | #8 | #2 | #4 | #5 | #6 | #7 |
| color (0=least green; 1=most green) | 0 | 0.2 | 1 | 0.2 | 0.4 | 0.5 | 1 | 1 |
| red bark just below twigs | 0 | 0 | 1 | 0 | 0 | 1 | 1 | 0 |
| twig reddish on one side | 1 | 1 | 0.8 | 0.5 | 1 | 0 | 0.2 | 1 |
| internode min (mm) | 10 | 10 | 3 | 6 | 2 | 2 | 2 | 3 |
| internode max (mm) | 25 | 18 | 8 | 17 | 17 | 4 | 6 | 13 |
| twig short hair mean orientation (90 = spreading; 180 = appressed toward tip) | 135 | 135 | 90 | 80 | 80 | 80 | 90 | 80 |
| twig short hair length (mm) | 0.1 | 0.2 | 0.1 | 0.2 | 0.05 | 0.2 | 0.1 | 0.1 |
| twig short hair density (0 = none; 1 = max in samples) | 0.5 | 0.5 | 0.4 | 0.5 | 0.1 | 0.8 | 0.3 | 0.3 |
| twig long hair # per internode min (count just ones seen from side in one view) | 28 | 25 | 5 | 3 | 3 | 0 | 0 | 0 |
| twig long hair # per internode max | 33 | 31 | 9 | 6 | 8 | 6 | 0 | 0 |
| twig long hair length min (mm) | 0.5 | 0.6 | 0.3 | 0.5 | 0.5 | 0.4 | 0.1 | 0.1 |
| twig long hair length max (mm) | 2 | 1.9 | 1.6 | 1 | 1.6 | 1 | 0.1 | 0.2 |
| largest leaf on twig petiole length (mm) | 10 | 10 | 9 | 9 | 9 | 8 | 8 | 10 |
| largest leaf blade length (mm) | 50 | 60 | 40 | 45 | 41 | 32 | 40 | 32 |
| largest leaf width (mm) | 28 | 39 | 22 | 24 | 22 | 19 | 26 | 22 |
| leaf shininess (0 to 1) | 0.2 | 1 | 1 | 1 | 0.2 | 1 | 1 | 1 |
| leaf blade hairiness (0=glabrous; 1 max hairiness in samples) | 1 | 1 | 0.1 | 0.3 | 0.1 | 0.1 | 0.2 | 0.4 |
| notes | no nascent infl | last year's stems | ||||||
The PCA results are shown in Fig. 2.
Fig. 2. PCA for all samples using all characteristics. The points are labeled separately for samples with abundant long hairs; some long hairs on any twig; and no long hairs at all.
The PCA shows that specimens with abundant long hairs, samples #1 and #3, are clearly separate from the other samples. These specimens correspond to ssp. mollis. Presumably, this means the other specimens are not ssp. mollis, despite the presence of some long hairs on some twigs of some of them.
The PCA1 values show that every characteristic except the twig short hair density and length contributed to the separation of ssp. mollis from the other specimens. Table 3 gives how important each characteristic was for PCA1, sorted in declining order of importance.
Table 3. How Important Each Characteristic Was for PCA 1
| PCA1 coefficient | Characteristic |
|---|---|
| 0.30 | twig long hair # per internode min (count just ones seen from side in one view) |
| 0.30 | twig long hair # per internode max |
| 0.30 | internode min (mm) |
| 0.29 | leaf blade hairiness (0=glabrous; 1 max hairiness in samples) |
| 0.28 | twig short hair mean orientation (90 = spreading; 180 = appressed toward tip) |
| 0.28 | largest leaf blade length (mm) |
| 0.28 | internode max (mm) |
| -0.25 | color |
| 0.25 | largest leaf width (mm) |
| 0.24 | twig long hair length max (mm) |
| 0.24 | largest leaf on twig petiole length (mm) |
| 0.23 | twig long hair length min (mm) |
| -0.21 | red bark just below twigs |
| 0.20 | twig reddish on one side |
| -0.14 | leaf shininess (0 to 1) |
| 0.05 | twig short hair length (mm) |
| 0.03 | twig short hair density (0 = none; 1 = max in samples) |
To see how well the samples would separate without using the characteristics of the long hairs, Tom ran another PCA without using any long hair characters. The PCA plot turns out to be nearly identical, showing that the long hairs are just the most obvious marker for ssp. mollis.
Another separate PCA, using just the specimens that were not mollis, did not result in any clearly separate groupings, implying either that all these samples are the same taxa, or that if they are not the same taxa, they are separated by characteristics not measured here.