Lecture #9 & 10 | Parsimony Cladistics

Internal nodes/External Branches and Nodes

Optimality Criterion (OC)

Uses optimality criteria to choose among set of possible trees

Advantage: OC methods require an explicit function for relating a tree to data

Disadvantage: Can be computationally expensive; the problem of finding the best tree from among all possible trees is difficult

Synapomorphy

Derived character state shared by two or more taxa and their common ancestor

Plesiomorphy

primitive or ancestral character state

Autapomorphy

Unique change not informative for relationships

Apomorphy

Any derived character state or feature within a lineage

Parisomony

The shortest hypothesis is the best, even though the alternative may be the correct hypothesis

Procedure or parsimony

1. selective suite of informative characters (from 1 to many)

2. code alternative character states

3. estimate based on the number of shared apomorphic states

4. beginning w/ the taxa that share the most derived character states, begin to add other taxa

How to select characters in parsimony

As many characters should be chosen

• each character gives equal weight to the determination of overall similarity

How to manually create a parsimony

1. group taxa with most synapomorphies

2. add the next group with most synapomorphies

3. add remaining taxon and map all characters

Fitch analysis

All character states unordered

Wagner analysis

All character ordered ( no difference if all binary)

General analysis

a mix of both ordered and unordered characters. Some characters may be better suited for ordering that others

How do fitch and Wagner analysis compare in the same data

Notice that with Wagner, that same data takes more steps

• Parsimony favors the tree that has the shortest length

Clarification of a “step” in cladistics

Each evolutionary novelty (apomorphy) is one step

• Includes:

  • homoplasies (convergence and parallelism)

  • reversals

  • autapomorphies

• Plesiomorphies and characters found in all taxa are not counted

Exhaustive search

A way to search for the most optimal tree by evaluating all trees

• should be done for 10 or few taxa

Exact search

A way to search for the most optimal tree where all trees are not evaluated, but search is guaranteed to find the best tree

• limited is less than 20 taxa

Uses a branch and bound algorithm that focuses on sets of trees

Heuristic searches

A way to search for the most optimal tree but is not guaranteed to find the best tree

• searches are almost always heuristic with more than 20 taxa

Steps:

1. Build initial search tree

2. swap branches/taxa to search for a shorter tree

3. search and keep all trees of the same length

Nearest neighbor interchange (NNI)

A method of branch swapping that swaps the closest branches only

• computationally fast

  

Subtree pruning and regrafting (SPR)

Intermediate changes grafting only at the point cut off

Tree bisection and reconnection (TBR)

Major changes involving cutting of groups and reattaching at any of the internal branch points

• it is the most computationally expensive, but best for making rearrangements more likely to find the shortest tree

• Most common method

Heuristic strategies

1. build initial search tree (asis, closest, simple, random)

2. swap branches/taxa to search for a shorter tree

3. search and keep all trees of the same length

Stepwise addition in Heuristics

1. Asis: order in the data matrix

2. closest: starts with shortest 3-taxon tree adds taxa in order that produces the least increase in tree length

3. simple: the first taxon in the matrix is the reference

   1. taxa are added to it in the order of their decreasing similarity to the reference

4. random: taxa are added in a a random sequence, many different sequences can be used

   

Consistency index (CI)

a measurement of tree strength that

equation: m/s

• m=minimum steps

• s=actual number of steps

Measures the amount of homoplasy on a particular tree

• decreases with the number of evolutionary steps needed to explain the distribution of characters

1: A character has perfect consistency

Two state characters (0/1): CI value of 0.5

Retention index (RI)

measurement of the fit of a character or how much a character change contributes to a tree topology

(g-s)/(g-m)

• m=minimum steps

• s=actual number of steps

• g= minimum number of any particular state on the tree

A character of perfect fit gets an RI of 1

An autapomorphy gets an RI of 0

Rescaled consistency index

ri x ci

A function to more adequately scale the value of the fit of a character (ri) to 0

• trees that have a retention index of less than 0.7 may have multiple fundamentally different tree topologies that can be of the same length, so rigorous analyses are needed

Parsimony support examples

Problems of cladistics methods

1. choice of characters may influence results

2. chance of unresolved or conflicting nodes on the tree

3. homoplasy (convergence, parallelism, and reversals (loss of acquired traits))

What happens if there are multiple trees that have the same “shortest” length

different trees can be combined together into a consensus tree

multiple due to

• alternative equally parsimonious optimizations of homoplastic characters

• missing data

  • or both

Most common relationships are summarizes with consensus trees

Strict consensus methods

only includes groups that appear in all of the input trees

• most conservative model for consensus trees

Majority rule consensus tree

tree that includes clades that appear in more than half of a collection of trees

• numbers indicate frequency of clades in the fundamental trees

Bootstrap

Commonly used measure of support for many of the OC methods of analysis

• characters are resampled randomly from original data, with replacement, until new data set with original number of observations obtained

  

• tree is computed for each replicated data set

• agreement among the resulting trees is summarized with a majority-rule consensus tree

• BP proportion: frequency of occurrence of a clade and is a measure of support for a group

Intersections (A ∩ B)

the same item found in each group

Note: work through problem in lecture 10

Union ( A ∪ B )

All items in each group

Note: work through problem in lecture 10

Advantages of parsimony

• simple method - easily understood operation

• does not depend on an explicit model of evolution

• gives both trees and associated hypotheses of character

  evolution

• should give reliable results if the data is well structured and

  homoplasy is either rare or widely (randomly) distributed on

  the tree

Disadvantages

• may give misleading results if homoplasy is common or concentrated in particular parts of the tree

• underestimates branch lengths for molecular data

• simple model of evolution, but is simple the best

• parsimony often justified on philosophical groups

• not compelling for molecular