02.05.2013 Views

Evolution__3rd_Edition

Evolution__3rd_Edition

Evolution__3rd_Edition

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

..<br />

(a) Phenetic measurements for five species<br />

Character 1<br />

(length of tibia)<br />

1 2<br />

A<br />

Character 2 (length of wing vein)<br />

(b) Nearest neighbor (c) Average neighbor<br />

1<br />

1<br />

Similarity can be measured for<br />

multiple phenetic characters<br />

2<br />

3<br />

4<br />

5<br />

3<br />

4<br />

B<br />

5<br />

2<br />

3<br />

4<br />

5<br />

CHAPTER 16 / Classification and <strong>Evolution</strong> 477<br />

Figure 16.2<br />

(a) The phenetic similarity between species can be expressed<br />

graphically. Suppose five species have been measured for two<br />

characters, for instance the length of a wing vein and the length<br />

of the tibia. The x-axis is the measurement for each species of the<br />

length of a wing vein, and the y-axis for the length of the tibia. The<br />

distance between two species on the graph is the phenetic distance<br />

between them. (Notice that the distance on the graph is different<br />

from the measure of mean character distance in Table 16.1.)<br />

(b) The phenetic classification by the nearest “nearest neighbor”<br />

technique puts species 3 with the group (cluster A) that has the<br />

nearest individual neighboring species (species 2). (c) But if it uses<br />

the nearest “average neighbor” technique it puts species 3 with the<br />

group (cluster B) that has the nearest average for all its species.<br />

Species 4–5 have a nearer average distance. (The average is simply<br />

the average of the distances of species 1–2 and of species 4–5 to<br />

species 3.)<br />

of phenetic characters, there is no way to decide which of the many classifications is<br />

the best.<br />

The next step is to define the classification not by a few characters but by many. This<br />

became possible in the 1950s and 1960s as the statistical and computational apparatus<br />

became available for aggregating large numbers of phenetic measures into one grand<br />

measure of phenetic similarity. The aim of the numerical phenetic school was to measure<br />

so many characters that the idiosyncracies of particular samples would disappear.<br />

The resulting classification groups the units according to their whole phenotype.<br />

How do we aggregate a large number of measures into a single combined measure<br />

of phenetic similarity? Several methods exist, and we can illustrate one of them in a<br />

graph. We start with the simple case of two characters, though the extension to further<br />

dimensions is easy. Suppose that we wish to classify a group of fly species, and we have<br />

measured two characters, such as the length of a certain wing vein, and the length of<br />

the tibia of the hindleg. The average for each species can be represented as a point<br />

(Figure 16.2). For any pair of species, the average difference in their wing vein lengths is<br />

the distance between their points on the x-axis, and the difference between the lengths<br />

of their tibiae is the distance on the y-axis. If we used either character by itself, the<br />

classifications would differ; species 1 and 3 for instance have identical tibial lengths, but<br />

different wing vein measurements. The aggregate difference for the two characters can<br />

be measured simply by the distance between the two species in the two-dimensional<br />

space. The species are then classified by putting each with the species, or group of<br />

species, that it has the shortest distance to.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!