Lecture 3: Phenetics and Phylogenetics

Study Notes

Three primary approaches for analysis: Distance or Algorithmic Methods (Phenetics), Parsimony Method (Cladistics), and Model-based Methods.
Distance methods include Nearest Neighbour and UPGMA, yielding a single 'correct' tree.
Parsimony produces multiple possible trees based on maximum parsimonious characteristics.
Model-based methods such as Maximum Likelihood and Bayesian Analysis rely on statistical models for phylogenetic reconstruction.

Also known as Numerical Taxonomy or Taxometrics, phenetics is based on overall similarity of taxa, derived from phenotypes.
It originated from Biological Statistics and aims to group taxa based on shared characteristics.

Increased data collection leads to more reliable results; more characters equate to more data.
All characters are treated with equal weight, assuming uniform importance across traits.
Similarity between OTUs (Operational Taxonomic Units) is assessed character-by-character.
Distinct groups, or phenons, emerge from correlated character sets.
Phylogenetic conclusions can be drawn from phenograms, adding objectivity to taxonomy.
Classifications are founded on phenetic similarity.

Requires at least 60 characters for statistically accurate results, avoiding unintentional biases.
Characters must be represented in a binary form (present/absent) to ensure clarity in data analysis.

A data matrix is created based on presence or absence of characters among different OTUs.
Similarity coefficients calculate the overall similarity between pairs of OTUs, leading to the formation of a similarity or dissimilarity matrix.

Clustering helps visualize similarity better than tabular data by transitioning to tree structures.
Common methods include Nearest Neighbour (single linkage clustering), where the two most similar OTUs are linked iteratively.
The process is heuristic, where all OTUs are connected based on descending similarity.

Unweighted Average Linkage Clustering (UPGMA) links OTUs with the highest similarity, treating new clusters as single OTUs for further calculations.
The distance between clusters is determined by averaging distances from all constituent OTUs.
These methods culminate in distance trees showing relationships based on overall similarity, often illustrated using Euclidean Distance to gauge character differences.