1. Maximum Parsimony

2. Character-state Approaches
- MP (Maximum Parsimony) and ML (Maximum Likelihood) are two major character-state/discrete approaches in constructing phylogenetic trees

3. MP Method
MP chooses the tree(s) that require the fewest evolutionary changes.
in MP analysis, synapomorphic characters provide the basis for clade identification.
autapomorphic characters do not contribute to the topology of MP tree(s).

4. MP Method tree 1 change 5 changes 1 3 A G G A A G G A G 2 4 A A G
For each site, the goal is to reconstruct the evolution of that site on a tree subject to the constraint of invoking the fewest possible evolutionary changes.
Taxon-1 ATATT
Taxon-2 ATCGT
Taxon-3 GCAGT
Taxon-4 GCCGT
tree
1 change
5 changes 1 3 A G G A A G G A G 2 4 A A G

5. MP Method Site 2 (1 step) T C T C C T Site 3 (2 steps) A A
Taxon-1 ATATT T
Taxon-2 ATCGT C
Taxon-3 GCAGT
Taxon-4 GCCGT T C C T
Site 3 (2 steps) A A
Site 4 (1 step) A A
Site 5 (No step) C T T C G T or A G G T T A G G T T C C C C

6. MP Method tree 1 3 ((1,2),(3,4)) 2 4 Sites tree 1 2 3 4 5 Total
Taxon-1 ATATT 1 3
Taxon-2 ATCGT
Taxon-3 GCAGT
((1,2),(3,4))
Taxon-4 GCCGT 2 4
Sites
tree
Total
((1,2),(3,4))
((1,3),(2,4))
((1,4),(2,3))
Changes required for each site to fit the three possible trees for 4 sequences

7. Search for Most Parsimonious Trees
- most parsimonious trees can be obtained by exhaustive or heuristic searches.
- exhaustive search will guarantee the most parsimonious trees but time consuming.
- heuristic search reduces time of searching for the most parsimonious trees. However, the trees obtained may be suboptimal.

8. Parsimony Algorithms
Parsimony approaches comprise a family of related methods with varying assumptions about how character-state transformation occurs.
Fitch Parsimony
Wagner Parsimony
Dollo Parsimony
Camin-Sokal Parsimony

9. Fitch Parsimony
allows free reversibility of character states in the tree, with changes in any direction equally likely.
characters may be binary or unordered multistate.
State-0ne
State-Two
State-Three

10. Wagner Parsimony
allows free reversibility of character states in the tree, with changes in either direction equally likely.
characters may be binary or ordered multistate, although transformations among multistate characters must occur through intervening states only.
State-0ne
State-Two
State-Three

11. Dollo Parsimony
Dollo optimization is introduced in order to accommodate evolutionary scenarios in which it is considered most plausible a priori that each apomorphic state could only have arisen once and that all homoplasy (reversion) must be accounted for by secondary loss.
multiple reversions to the ancestral condition are allowed.
appropriate when probabilities of change among character states are highly asymmetric. eg. loss of a particular restriction site might be more likely than its gain.

12. Comparison Dollo - Fitch
Reversal
Apomorphic
Fitch optimization
Dollo optimization

13. Camin-Sokal Parsimony
assumption that all evolutionary changes are irreversible.
goes beyond Dollo by disallowing reversions to the ancestral condition.
the optimization is not employed widely with genetic data because most molecular characters probably violate this assumption.

14. Long Branch AttractionD B
- The edges leading to sequences/taxa A and C are long relative to other branches in the tree, reflecting the relatively greater number of substitutions that have occurred along those two edges.
- the long branch attraction occurs when rates of evolution show considerable variation among sequences, or where the sequences being analysed are quite divergent.

15. Long Branch AttractionD B
If the internal edge is short relative to the long terminal edges, then by chance alone A and C may acquire the same nucleotide independently. These convergences may outweigh the sites changing along the internal edge, and hence by the parsimony criterion the tree ((A,C),(B,D)) would be flavoured.
How to overcome this?

16. Long Branch AttractionD B
How to overcome Long Branch Attraction?
To reduce the effects of long edges is to add sequences/taxa that join onto those edges thus breaking them up.