08 Numerical Phenetic
08 Numerical Phenetic
08 Numerical Phenetic
Approach
Numerical Phenetics
• similarity only, not evolution, not phylogeny (1950-1960)
– computerized, reproducible, “objective,” modern
– goal to remove all subjectivity, measure everything, use
matrix of distances
– sometimes called “Numerical Taxonomy” after book by
Sokal & Sneath (1963)
– cluster based on overall similarity
– the end product of clustering process is a phenogram
not a classification or evolutionary trees
BIOSIST2010-07
OTU and phenetic criteria
• OTU (Operational Taxonomic Unit) refers to a very
heterogeneous class of entities. Some OTUs are
individuals, some are populations, some are historical
entities.
• Phenetic employ four criteria to determine whether or
not a given classification satisfactorily represents the
probable relationship of the included taxa :
1. Closeness to the original similarity matrix
2. Maximal similarity within the taxa
3. Maximal predictive value
4. Stability of classification
BIOSIST2010-07
Phenetics - the good
1. Phenetics deserves credit for having introduced the
computer into taxonomy
2. Quite useful in classifying large genera with numerous
closely related species, all with similar evolutionary
propensities and hence abundant homoplasy
3. Sometimes useful when a determination of the
degree of similarity is the objective
4. Demanded explicit character analysis & laid
groundwork for numerical phylogenetics
5. Still used for lower-level problems (e.g.
morphometrics, species demarcations) and some
higher-level problems
BIOSIST2010-07
Phenetic Problems and Weakness
1. Discrepancy between the endeavor to study overall
similarity and the actual use of a very limited number
of characters
2. Reproducible but no agreement on which clustering
algorithm or statistic to use…
- different algorithms = different trees
e.g. UPGMA, PCA… etc.
- although there is only one true evolutionary
tree, there are many alternative ‘similarity’
trees depending on the data & algorithm used
3. Similarity doesn’t always = evolutionary relationship!
BIOSIST2010-07
Weak Points of Phenetic Methods
1. Practical difficulties
a. The method is rather laborious and data assembly of a
complete character matrix for a large number of characters
and species is forbiddingly time consuming compared with
some other method
b. The success of the method in traditional taxonomy depends
on the availability of a large assortment of variable
morphological characters
c. Phenetic classification cannot be improved gradually, because
any addition of new characters requires a new analysis
d. Numerical phenetic has failed to make a substantial
contribution to the classification of any mature group or to
the classification of haigher taxa
BIOSIST2010-07
2. Inability to meet the claims of objectivity and repeatability
a. Subjectivity
b. Lack of theoretical justification of choice among clustering methods
c. Repeatability and choice of computational method
d. Repeatability and selection characters
BIOSIST2010-07
Phenetic Method
• Similarity coefficients and distances
– Imagine we are studying two animals A and B , by
identifying three characters a, b, c. which can be
given a numerical value
(For instance leg length; legs may be absent (=0) 0-
1cm long (=1) 1-2cm long (=2) etc.):
Character a Character b Character c
Organism A 1 2 3
Organism B 2 2 3
BIOSIST2010-07
• Similarity coefficient S
• Distance coefficient E
* May be measured, for example by the Euclidean
distance
* Where n is the number of characters
BIOSIST2010-07
• Examples
Below are reproduced two character matrices based upon fruit
and toys. The fruit character matrix is worked to show a table of
distances and a phenogram indicating the pattern of similarities
between fruits. A phenogram is a table of similarities.
• Classification of fruit
BIOSIST2010-07
BIOSIST2010-07
Numerical Phenetic Today and in the
Future
• Modern applications of phenetics are common in botany,
and some examples can be found in most issues of the
journal Systematic Botany. Indeed, due to the effects of
horizontal gene transfer, polyploid complexes and other
peculiarities of plant genomics, phenetic techniques in
botany - though less informative altogether - are also less
prone to errors compared cladistic analysis of DNA
sequences.
• In addition, many of the techniques developed by
phenetic taxonomists have been adopted and extended
by community ecologists, due to a similar need to deal
with large amounts of data.
BIOSIST2010-07
Reading sources
• Jones, S.B. and A.E. Luchsinger. 1987. Plant
Systematics. 2nd edition. McGraw Hill Book Co.
Singapore.
• Mayr, E. and Ashlock. 1991. Principles of
Systematic Zoology. McGraw Hill, Inc. United
States. pp 195-205.
• Ross, H. 1974. Biological Systematic. Addison-
Wesley Publ. Co., Inc. Philippines.
• http://homepage.mac.com/wis/Personal/lectur
es/human-evol/2.html
BIOSIST2010-07