Lecture #8 | Tree Building

Discrete characters

No overlapping variation

• teeth, no teeth

Divided into binary or multistate

• Binary (0 or 1)

• Multistate ( >2 states, 0 or 1 or 2)

Continuous characters

Measurements

Multistate characters

May be ordered (linear) or unordered

How are numerical scores assigned?

Usually (not always), the character state considered to be most ancestral is given the lowest numerical value, with more derived states given an increasingly higher value (1,2…) depending on the type of character

• usually assigned with an initial hypothesis of character change developed by the investigator

• Subject to reevaluation

Character polarity

Assignment of character order

• evolutionary history of a trait or feature of an organism

Outgroup method

A method to determine polarity

• If a character has 2 or more states, the state found in the next most related group (outgroup) is ancestral

• 

• By far the most common tree rooting method

• Best to have two outgroups because it defends against autapomorphy (unique change in character state that is not informative for relationships)

What is important when considering methods of analysis for a tree?

1. Informative-organized

2. Predictive-Provides information that is fundamental to the relationships

3. Stable- stable to new info from new taxa and data

4. Operational- based on a set of procedures that can be accepted and refuted

Newick Format

Method to represent graphical trees, with or without branch lengths, using parentheses, commas, and a semicolon

How to construct a newick tree

1. Identify internal nodes

   1. 

2. Add a set of parentheses for each internal node and a comma between the left and right descendants of each nodes. Add a semicolon to the end of the Newick tree

   

3. Each branch length is preceded by a colon (;)

4. Branch lengths (with their preceding :) are placed after the taxon names and after right parentheses (except the last one)

• Binomial names require apostrophes or underscores to link together

  

Unrooted trees

Lack temporal polarization

• point of common ancestry is missing

Rooted trees

Temporally polarized

• point of common interest is given

Determining unrooted number of taxa

where n = number of taxa

Determining number of rooted trees

Gene duplications root

Paralogous gene duplications that predate the common ancestor of a taxonomic group are used to root the tree

• root is placed between paralogous gene copies

  

Midpoint rooting

Tree is rooted on the midpoint between the two most distant leaves

• choose the midpoint between the two most distant external nodes

• assumes the rate of evolution is the same on the longest branches of the tree

Desirable properties of tree building methods

• Consistency: will the method converge on the correct solution given enough data

• Efficiency: How fast is the method

• Power: How much data is needed for a reasonable result

• Robustness: Will minor violations of the assumptions result in poor estimates of phylogeny

Types of data: Discrete versus Continuous

Discrete data is more common-few methods can handle continuous data

Types of data: Character versus Distance Data

Important that character comparisons between taxa can be used to develop distance matches, but reverse cannot happen

Types of data: Correct versus uncorrected data

Morphological characters may be standardized so that they all have equal value in an analysis

Cluster analysis

The recognition of groups of individuals on the basis of multiple characters. Groups may be mutually exclusive, hierarchic, or partially overlapping

Phonetics

classification based on numerous precisely delimited characters of equal weight and their comparison by an explicit method of grouping

Key points of cluster analysis

1. Objectivity

2. Polythetic Taxa: Groups based on character combinations

3. Many characters- use as many characters as possible

4. Equal weighting-every character has equal weight

5. Overall similarity: groups recognized on basis of overall similarity nothing else

6. Defining character polarity is nor important

How to create a cluster analysis

1. Select taxa that represents both the entire geographical range and the entire morphological range of variation

2. Select characters: As many characters as possible should be chosen. Each character gives equal weight to the determination of overall similarity

3. Calculation of similarity/dissimilarity matrix

   

4. Grouping OTUs by single linking (nearest neighbor)

Advantaged of cluster methods for determining relationships

1. Operational: clearly defined procedures

2. Communicable: anyone can code for characters and produce a classification without prior knowledge

3. No weighting or preference for certain characters

Problems of cluster methods

Relationships depicted are strongly affected by

1. choice of characters

2. number of taxa

3. type of similarity coefficient

4. Clustering technique applied to similarity matrix

5. higher categories are subjective

Groupings are more technique dependent providing an artificial grouping of taxa rather than moving towards a system of uncovering stable relationships

How often are phenetic (cluster) methods used?

Rarely do studies that utilize morphological characters ever use phenetic methods.

• However, phenetics are still used for relatively simple organisms like prokaryotes

What assumptions allow for use of distance methods for molecular data?

1. Molecular clock is assumed whereby mutations at any particular site in the genome are random and occur with equal frequency over time

2. most changes are observed

3. Character system is enormous with a potential to use the entire genome for analysis

4. Changes in the genome are expected to be independent of environmental or selective pressure and less subject to convergence

But we know

1. There is a preference for transitions

2. not all positions of codons change at equal rates

3. different regions of the genome,e are more conservative

Morphometrics

The quantitative description, analysis, and interpretation of shape and shape variation in biology

Single linkage clustering method

UPGMA- Unweighted pair group mean average

Neighbor-Joining Methods

• Widely used for constructing phylogenetic trees with molecular data

• Can be applied to the data for the corrected matrix conversion

• Assumes additivity, not ultrametricity, so all branch length divergencies are not necessarily equal

  • Branch lengths in the matrix and the tree path length match perfectly and there is a single and unique additive tree that fits the distance matrix

    

Advantaged of NJ

• Branch lengths are additive and reflect the true distances between taxa

• Fast computational time

• Can invoke outgrip rooting of the tree

• Can empty various models of character state evolution to adjust branch lengths relationships

Disadvantages of NJ

• not possible to infer or directly map character back onto topology

• produces a single tree with no evaluation of competing hypotheses

• can produce a quick and dirty tree that may be very different from OC method trees