Please enable JavaScript.
Coggle requires JavaScript to display documents.
Inferring Phylogeny (Building Trees using Parsimony (Fitch Algorithm (What…
Inferring Phylogeny
Building Trees using
Parsimony
What is it?
-the best phylogeny is one that both
explains observed character data
and posits the
fewest evolutionary changes
How does it work?
-using
Fitch algorithm
to find the
minimum character state
for that tree, and define that number as the
parsimony score
for that particular tree.
-look at various possible trees and select the one with the
lowest parsimony score
(however, some trees could be tied for minimum parsimony score)
Fitch Algorithm
What is it?
tells us the minimum number of character changes that are needed for a given tree
How does it work?
starting at the tips of the branches and working to the roots:
-if daughter branches have
some/all same characters,
assign internal node as the
intersection
of the characters and do not increase tally.
-if daughter branches have
no same characters
, assign internal node as
union
of the characters and increase tally by 1.
To note:
The algorithm does
not
tell us the most likely
character states
for each
ancestral node
. The process of assigning states to interior nodes is a way to
count number of changes
,
not to reconstruct ancestral types.
Limitation
:
long branch attraction
; since evolutionary changes occur at different rates on different branches, parsimony may incorrectly infer too close a relationship between rapidly evolving branches.
Rooting Trees
parsimony
is
unable
to
distinguish
between multiple rooted trees that are derived from the same unrooted tree, since the parsimony count is the same for all these rooted trees.
How, then, do we root a tree?
-Using an
outgroup
(a related population that branched off earlier in evolutionary history)
-The outgroup method: select the
branch
on the unrooted tree leading to the
outgroup
and
draw a tree around a point
on that branch to produce a rooted tree.
What can rooted trees tell us?
-
Polarity
of characters (order of appearance of characters in evolutionary time)
-
Phylogeography
(knowing the root of the tree can tell us the story of how a species moved across the globe)
Building Trees using
Distance Methods
Method
: measure
pairwise distances between species
and use these distances to reconstruct a tree.
Aim
: find a tree such that
distance along the branches
between any two species is
equal
to the
distance that we measured
between those two species
Step 1: Measuring Distance btwn Species
-using DNA/amino acid sequences lined up and count the number of base pair/amino acid differences
-using Sequence Alignment
(but it has its limitations when there are multiple insertions and deletions
)
Step 2: Constructing a Tree from Distance Measurements
-form a distance matrix (a table that lists distance btwn each species pair)
-using different methods to obtain the tree from the distance measurements
Problems with Distance Methods
-
phenetic
in nature: groups species together based on similarity
without
reflecting any
evolutionary relationship/history
-assume that the similarities are
homologies
, not
analogies
-greater number of differences may not be due to distantly related species, but could be due to
faster evolving species
Phylogenies and
Statistical Confidence
How strongly does the character data support a given phylogeny?
-test this by using 2 techniques to quantify this statistical confidence:
bootstrap resampling & odds ratio testing
-usually focus on making statements of confidence on
features
of the tree (focus on a
single monophyletic clade
)
Bootstrap Resampling
Construct a
character matrix
for traits and species
Resample
from original matrix to create
bootstrap replicate data sets
Build
trees
from these replicate data sets to produce
bootstrap replicate phylogenies
If species form a monophyletic clade in 90% of replicate phylogenies, the clade has
90% bootstrap support.
Odds Ratio Testing
Comparing
best possible tree
where clade A is
monophyletic
against
best possible tree
where clade A is
not monophyletic
Then see how much better the former is from the latter
Evidence
used for Reconstructing Phylogenetic Trees
Fossil Record
Common Descent and Phylogeography
using fossil records, we can map absolute
time
and
place
onto phylogenies
can tell us about
micro
- (migration and dispersal) and
macroevolutionary
events (adaptive radiation and extinction)
linked to
continental drift
(common ancestor and after continents split up, the species diverged according to different environments)
Phylogeny and
Comparative Method
compare
traits
across diff species, can
identify trends/patterns
and see if evolutionary events (eg. NS) produces specific relationships btwn certain traits. See if traits
evolve together
have to consider
phylogenetic
relationships
btwn the species you are comparing. Not doing so may produce
misleading estimates
of the
significance
of patterns observed
Must take into account that
data points
for traits are
not independent observations
, but instead are
LINKED
up by shared evolutionary history
How do we tackle this problem?
Using
Independent Contrasts
Don't look at absolute character states, but rather
difference
in character states btwn each pair that we are considering
Eg. difference in testes mass and difference in reproductive age.
Pair
of differences become the
first data point
after accumulating a
set of independent contrasts
(data points are now independent of each other), can plot a
linear regression
to ascertain
relationship btwn the traits