slide1 l.
Download
Skip this Video
Loading SlideShow in 5 Seconds..
Section 9 Resolving Taxonomic Uncertainties & Defining Management Units PowerPoint Presentation
Download Presentation
Section 9 Resolving Taxonomic Uncertainties & Defining Management Units

Loading in 2 Seconds...

play fullscreen
1 / 110

Section 9 Resolving Taxonomic Uncertainties & Defining Management Units - PowerPoint PPT Presentation


  • 219 Views
  • Uploaded on

Section 9 Resolving Taxonomic Uncertainties & Defining Management Units. The taxonomic status of, and relationships among many taxa are unresolved. In conservation, many erroneous decisions may result if the taxonomic status of populations is not correctly assigned, such as:.

loader
I am the owner, or an agent authorized to act on behalf of the owner, of the copyrighted work described.
capcha
Download Presentation

PowerPoint Slideshow about 'Section 9 Resolving Taxonomic Uncertainties & Defining Management Units' - doane


An Image/Link below is provided (as is) to download presentation

Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author.While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server.


- - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - -
Presentation Transcript
slide1

Section 9

Resolving Taxonomic Uncertainties &

Defining Management Units

The taxonomic status of, and relationships among

many taxa are unresolved.

In conservation, many erroneous decisions may

result if the taxonomic status of populations is

not correctly assigned, such as:

slide2

Unrecognized endangered species may be allowed

to become extinct.

Incorrectly diagnosed species may be hybridized

with other species, resulting in reduced

reproductive fitness.

Resources may be wasted on abundant species,

or hybrid populations.

slide3

Populations that could be used to improve the

fitness of inbred populations may be overlooked.

Endangered species may be denied legal protection

while populations of common species, or hybrids

between species, may be granted protection.

slide4

Recent molecular studies among sea turtles

compared the Kemp’s ridley turtle (Lepidochelys

kempi) and the similar olive ridley turtle (L. olivacea)

and supported recognition of Kemp’s ridley turtle

as a valid species.

Studies of the genetics of minke whales

(Balaenoptera acutorostrata) have led investigators

to advocate that the Northern and Southern

Hemisphere populations be treated as two distinct

species (Hoelzel and Dover, 1991).

slide5

Similar conclusions were

reached based on

molecular studies of

sympatric populations of

killer whales

(Orcinus orca).

This case is particularly interesting because it

suggests that observed differences in behavior

in sympartric populations, so called “resource

polymorphisms”, may be genetically based

(Hoelzel 1998).

slide6

Proper identification and determination of

evolutionary relationships can prevent

hybridization, and sometimes genetic extinction

of “look-alike” species.

slide7

Case of the Extinct Dusky Sea Side Sparrow

1872, a melanistic form of

seaside sparrow was discovered

in Brevard co. FL and described

as a distinct species:

Ammodramus nigrescens.

1960s the population (now a

subspecies) was in severe

decline due to habitat alterations

and the Dusky Seaside Sparrow was placed on the

U.S. Endangered Species List.

slide8

Currently, about 9 subspecies recognized

with more or less abutting ranges

along the species coastal-marsh

habitat from New England to

south Texas.

1980, the few remaining

birds (all males) were

brought into captivity

and mated to individuals

from a Gulf Coast population.

slide9

The objective was to produce F1 hybrids and then

backcross progeny (the latter carrying primarily

dusky nuclear genes) for eventual reintroduction.

The breeding program was not successful and thus

discontinued.

Avis and Nelson (1989) assayed mtDNA haplotypes

from 40 seaside sparrows representing 7 named

subspecies and the last available dusky male, which

died in captivity in 1987.

slide10

Gulf Coast

haplotypes

Atlantic Coast

haplotypes

7

5

8

3

1

6

11

2

9

4

10

Most common

Atlantic haplotype

and haplotype of

Dusky seaside

sparrow

slide11

Thus, the traditional taxonomy for seaside

sparrows, from which conservation priorities

were derived, probably had been a misleading

guide to evolutionary relationships in this complex

for two reasons:

in failure to recognize the fundamental

phylogenetic dichotomy between Atlantic & Gulf

populations.

2. in taxonomic emphasis on distinctions within

both coastal regions that appear evolutionarily minor

compared to the between-region genetic differences.

slide12

Captive breeding of

gazelles and dik-diks that

were supposedly of the same

species ha sometimes produced

infertile offspring.

Subsequent cytogenetic

analyses revealed that the

parents were of different

species.

slide13

Dealing specifically with dik-diks, Benirschke &

Kumamoto (1991) noted that not only were

individuals of different species bred together

in captivity, but also hybrids of Kirk’s (Madoqua

kirkii) and Guenther’s (M. rhyncotragus) dik-diks

were found in 300 collections.

These authors concluded that a cytogenetic

analysis should be mandatory prior to captive

breeding populations are established to eliminate

unnecessary hybridization and reduced fertility.

slide14

Resolving Taxonomic

Uncertainties

slide15

Phylogenetic trees are used to resolve taxonomic

uncertainties.

A phylogenetic tree is composed of lines called

branches that intersect and terminate at nodes.

The nodes at the tips of the branches represent

the taxa that exist today and that we can

actually examine.

The internal nodes represent ancestral taxa,

whose properties we can only infer from the

existing data.

slide16

Rooted tree whose branch tips

represent 5 taxa (A - E) in a

clade, with 4 internal nodes

(R, X, Y, Z) representing

ancestral taxa, including the

root (R).

The numbers on branches indicate the

number of changes in a particular sequence

that occurred along that branch.

These numbers represent the branch lengths.

A

2

1

Z

B

1

2

C

3

1

1

Y

D

R

7

E

slide17

Even if exact values are not

provided, the relative lengths

of the branches may be drawn

in proportion to the number of

changes along that branch.

This tree is additive because the

distance between any two nodes

equals the sum of the lengths of all

branches between them.

If multiple substitutions have occurred at any site,

then additivity will not hold unless distances are

corrected for multiple substitutions.

A

2

1

Z

B

1

2

C

3

1

1

Y

D

R

7

E

slide18

A tree is said to be rooted if

there is a particular node -- the

root -- from which a unique

directional path leads to each

extant taxon.

In this tree, R is the root

because it is the only internal

node from which all other nodes

can be reached by moving forward

(toward the tips).

The root is the common ancestor of all taxa in the

analysis.

A

2

1

Z

B

1

2

C

3

1

1

Y

D

R

7

E

slide19

Unrooted tree, such as these,

specify only the relationships

among the taxa, and DO NOT

define evolutionary pathways.

For 4 taxa, there are only 3

possible unrooted trees.

Once a root is identified, 5

different rooted trees

can be created for EACH of

these unrooted trees, each with a distinctive

branching pattern reflecting a different evolutionary

history.

C

A

D

B

B

A

D

C

C

A

B

D

slide20

The number of possible trees, both rooted and

unrooted, increases dramatically as the number of

taxa increases.

Let s be the number of taxa, the number of possible

unrooted trees is:

(2s - 5)! / [2s-3(s-3)!]

the number of possible rooted trees is:

(2s - 3)! / [2s-3(s-3)!]

slide21

Taxa unrooted rooted

trees trees

4 3 15

8 13,395 135,135

10 2,027,025 34,459,425

22 1 x 1023almost a mole

of trees

50 3 x 1074more trees

than atoms in

the universe

slide22

Unrooted trees tell us only about phylogenetic

relationships; they tell us nothing about the

directions of evolution -- the order of descent.

Rooted trees tell us about the order of descent

from the root toward the tips of the tree.

While unrooted trees are always more “correct”

in that they don’t imply knowledge that we do

not have, they are considerably less

informative.

slide23

Alignment of DNA sequences

A pair of sequences can be aligned by writing one

above the other in such a way as to maximize the

number of residues that match by introducing gaps

into one or the other sequence.

Biologically, these gaps are assumed to represent

insertions or deletions that occurred as the

sequences diverged from a common ancestor.

slide24

If we could insert as many gaps as we chose, we

could align any two random, unrelated sequences so

that all residues either matched perfectly or were

across from a gap in the other sequence.

Such an alignment would be meaningless!!!

It is necessary to somehow constrain the number

of gaps so that the resulting alignment makes

biological sense.

slide25

To do this, a scoring system is used so that

matching residues get some sort of positive

numerical score, and gaps get some sort of negative

score, or gap penalty.

An alignment program seeks an arrangement that

maximizes the net score.

For nucleic acid alignments, matching residues

usually get a score of 1 and mismatches get a score

of 0.

slide26

Gap penalties are typically set by the user and

typically there is a penalty for creating a gap

plus an extra penalty for the length of the gap.

Aligning a pair of sequences is not a computationally

difficult process, and a variety of programs exist

to align sequence pairs.

Multiple alignments are considerably more complex,

and only a few programs do a really good job.

CLUSTALX is one of the best tools for creating

multiple sequence alignments.

slide27

An alignment is not an absolute thing.

It is a “best guess” according to some algorithm

used by a computer program.

One cannot simply have a program compute an

alignment and, without further thought, use

that alignment to create a phylogeny.

slide28

Distance Based Methods of Tree Construction

In these methods, distances are expressed as the

fraction of sites that differ between 2 sequences

in a multiple alignment.

It is fairly obvious that a pair of sequences differing

at only 10% of their sites are more closely related

than a pair differing at 30% of their sites.

It also makes sense that the more time has passed

since two sequences diverged from a common

ancestor, the more the sequences will differ.

slide29

Although the latter assumption is reasonable, it

is not always true.

It might be untrue because one lineage evolved

faster than the other.

Even if two lineages evolved at the same rate,

the assumption might be untrue because of

multiple substitutions.

slide30

As two sequences diverge from a common

ancestor, each nucleotide substitution initially

will increase the number of differences between

the two lineages.

As those differences accumulate, however, it

becomes increasingly likely that a substitution

will occur at the same site where an earlier

substitution occurred.

slide31

While there are statistical corrections used to

estimate corrected distances from the number of

observed differences, differences almost always

underestimate the actual amount of change along

lineages.

The two most popular distance methods, UPGMA

and Neighbor-Joining, are both algorithmic

methods -- i.e., they use a specific series of

calculations to estimate a tree.

slide32

The calculations involve manipulations of a distance

matrix that is derived from a multiple alignment.

Starting with the multiple alignment, both programs

calculate for each pair of taxa the distance, or the

fraction of differences, between the two sequences

and write that distance to a matrix.

slide33

UPGMA:

UPGMA (Unweighted Pair-Group Method with

Arithmetic Mean) is an example of a clustering

method.

We covered this procedure in chapter 13.

UPGMA has built into it an assumption that the

tree is additive and that it is ultrametric -- all

taxa are equally distant from a root -- an assumption

that is very unlikely. For that and other reasons,

UPGMA is rarely used today.

slide34

Neighbor-Joining (NJ):

NJ is similar to UPGMA in that it manipulates a

distance matrix, reducing it in size at each step,

then reconstructs the tree from that series of

matrices.

It differs from UPGMA in that it does not construct

clusters but directly calculates distances to internal

nodes.

slide35

From the original matrix, NJ first calculates for

each taxon its net divergence from all other taxa as

the sum of the individual distances from the

taxon.

It then uses the net divergence to calculate a

corrected distance matrix.

NJ then finds the pair of taxa with the lowest

corrected distance and calculates the distance

from each of those taxa to the node that joins

them.

slide36

A new matrix is then created in which the new

node is substituted for those two taxa.

NJ does not assume that all taxa are

equidistant from a root.

NJ is, like parsimony, a minimum-change method,

but it does not guarantee finding the tree with

the smallest overall distance.

slide37

Indeed, there are cases in which many shorter

trees than the NJ exist.

Some authors think that the best use of an NJ

tree is as a starting point for a model-based

analysis such as Maximum-likelihood.

slide38

Parsimony:

Parsimony is based on the assumption that the

most likely tree is the one that requires the fewest

number of changes to explain the data.

The basic premise of parsimony is that taxa

sharing a common characteristic do so because

they inherited that characteristic from a common

ancestor.

slide39

When conflict occur, they are explained by

reversal, convergence, or parallelism and these

explanations are gathered under the term

homoplasy.

Homoplasies are regarded as “extra” steps or

hypotheses that are required to explain the data.

Parsimony operates by selecting the tree or trees

that minimize the number of evolutionary steps,

including homoplasies, required to explain the data.

slide40

Parsimony or minimum change, is the criterion for

choosing the best tree.

For protein or nucleotide sequences, the data are

aligned sequences.

Each site in each alignment is a character, and each

character can have a different state in different

taxa.

Not all characters are useful in constructing a

parsimony tree.

slide41

Invariant characters, those that have the same

state in all taxa, are obviously useless and are

ignored by parsimony.

Also ignored are characters in which a state occurs

in only on taxon.

An algorithm is used to determine the minimum

number of steps necessary for any given tree to be

consistent with the data.

That number is the score for the tree, and the tree

or trees with the lowest score are most parsimonious.

slide42

The algorithm is used to evaluate a possible tree

at each informative site.

Consider a set of 6 taxa, named 1 -- 6.

At some site (character) in the alignment, the

states of that character are:

1 = A

2 = C

3 = A

4 = G

5 = G

6 = C

slide43

There are 105 possible unrooted trees of 6 taxa.

We will pick one unrooted tree, but all will be

evaluated by the computer.

If we root this tree at taxon 1, we get the

following tree:

C

2

A

3

G

4

A

1

G

5

C

6

slide44

6C

5G

3A

4G

W

X

2C

Y

Z

1A

The algorithm starts at a tip and moves to the

interior node that connects to another tip.

If the two tips have the same state, they assign

that state to the node; if they do not, they assign

an “or” state.

slide45

6C

5G

3A

4G

W

X

2C

Y

Z

1A

Thus, node W is assigned the state A or G, and

node X the state G or C. Node Y connects nodes

W and X. Because the states at nodes W and X

both include G, node Y is assigned the state G.

Node Z is assigned the state C or G as follows:

slide46

6C

5G

3A

4G

G or C

A or G

2C

G

C or G

1A

Once the root has been reached, the algorithm

proceeds back up from the root toward the tips.

Because node Z does not include the state at the

node that is ancestral to it (taxon 1), its assignment

is arbitrary

slide47

6C

5G

6C

3A

5G

4G

3A

4G

G

G

2C

G or C

A or G

2C

G

G

G

1A

C or G

1A

Assume that it is assigned state G. Node Y is already

assigned, so the algorithm moves to node W. Node

W is assigned G because that assignment does not

require a change from the node that is ancestral

to it. Similarly, node X is assigned state G.

slide48

Each branch along which the

state changed, indicated by

thick branches, is counted.

This tree has 4 changes.

The other possible rootings of the tree are

considered in the same way, and if a different

rooting of the tree produces fewer changes, that is

the score for that site.

5G

3A

4G

G

G

2C

G

G

1A

slide49

The parsimony program evaluates the tree for

each informative site, then adds up the changes

to calculate the minimum number of changes for

that particular tree.

As it works its way through the various possible

trees, the program keeps track of the tree

(or trees) with the lowest scores.

slide50

Evolutionary Models:

Sequences diverge from a common ancestor because

mutations occur and some fraction of those

mutations are fixed into the evolving population by

selection and by chance, resulting in the

substitution of one nucleotide for another at

various sites.

To reconstruct evolutionary trees, we must make

some assumptions about the substitution process

and state those assumptions in the form of a model.

slide51

The simplest model is one in which the probabilities

of any nucleotide changing to any other nucleotide

are equal.

To predict the probability that a particular

nucleotide at a particular site will change to some

other specific nucleotide over some time interval,

we need to know the instantaneous rate of change.

This simple model has only one parameter and isknown as the Jukes-Cantor model.

slide52

If we know there is a G at some site at t = 0, we

can ask what is the probability that there will still

be a G at that site at some time t, and what is the

probability that there will be, for instance, an A at

that site instead.

These are expressed, respectively, as P(GG)(t) and

P(GA)(t). If the substitution rate is  per time

unit, then:

P(GG)(t) = 1/4 + 3/4e-4 t and P(GA)(t)=1/4-1/ 4e-4 t

slide53

Because according the the Juke-Cantor model all

substitutions are equally likely, a more general

statement is:

P(ii)(t) = 1/4 + 3/4e-4 t and P(ij)(t)=1/4-1/ 4e-4 t

When t is very close to zero, the probability that

the site has not changed, P(ii), is very close to 1,

while P(ij) the probability that the nucleotide at

that site has changed from i to some other

nucleotide, j -- is close to 0.

slide54

As time goes on, both probabilities approach 0.25;

the time required for that approach depends on

.

We can construct a table that shows the

instantaneous rates for each of the possibilities

for change at a site as:

slide55

Substituted Base

A C G T

A -3   

Original base C  -3   

G   -3  

T    -3 

This matrix is commonly called the Q-matrix.

This is not a matrix of probabilities but a matrix of

rates, and the elements in a row sum to 0.

slide56

The Jukes-Cantor model is the simplest, but not

very realistic.

We know that not all changes occur at the same

rate and a variety of models have been proposed

that allow the specification of different rates.

The most general is one in which each different

substitution can occur at a different rate, which

depends upon the equilibrium frequency of that

nucleotide, symbolized as A for the equilibrium

frequency of A.

slide57

General Nonreversible Model

-aA-b G-c t a C b G c T

d A-d A-e G-f T e G f T

Q = g A h C-g A-h C-i  Ti T

j A k C l G-j A-k C-l G

All other important models are special cases of

this general nonreversible model.

In Kimura’s two-parameter model, transitions

occur at one rate, , and transversions occur at a

different rate, .

slide58

Kimura’s Two-Parameter Model

-a-2b b a b

b -a-2b b a

Q = a b -a-2b b

b a b -a-2b

slide59

In the General Time-Reversible (GTR) model, there

are 6 different rates. Time-reversible models

assume that the overall instantaneous rate of

change from base i to base j is the same as from

base j to base i.

-aA-b G-c t a C b G c T

a A-a A-d G-e T d G e T

Q = b A d C-b A-d C-f  Tf T

c A e C f G-c A-e C-f G

slide60

When these evolutionary models are used to

reconstruct trees, one may either assign specific

values to those rates, or estimate the values from

the data.

These models implicitly assume that the rates are

the same at all sites.

It is also possible to include rate variation across

sites in the models.

slide61

Maximum Likelihood

Maximum likelihood (ML) tries to infer an

evolutionary tree by finding that tree that

maximizes the probability of observing the data.

For sequences, the data is the alignment of

nucleotides or amino acids.

slide62

TCAAAAATGGCTTTATTCGCTTAATGCCGTTA

TCCGTGATGGATTTATTTCTGCAATGCCTGTC

TTCGTGATGGATTTATTGCTGGTATGCCAGTC

TTCGTGACGGGTTTATCTCGGCAATGCCGGTC

We begin with an evolutionary model that gives the

instantaneous rates at which each of the 4 possible

nucleotides changes to each of the other 3 possible

nucleotides and a hypothetical tree of some topology

and with branches of some length.

There are three possible unrooted trees for 4 taxa,

one of which looks like the following for the site

in red:

slide63

T

C

Y

X

T

G

If the model is time-reversible, we can root the

tree at any node. One possible rooted tree is:

G

C

T

T

Y

X

slide64

We do not know the nucleotides at nodes X and Y,

but since there are four possibilities for X and

four for Y, there are 16 possible scenarios that

might lead to the previous tree, one of which is:

G

C

T

T

T

A

slide65

The probability of this scenario

is the probability of observing an

A at the root (PA), which might be

1/4 or might be the overall frequency

of A, depending on the model, time and the

probability of each change along the branches

leading to the tips.

The probability of changing from an A at the root

to a G at the tip is calculated from the

instantaneous rate matrix in the chosen model and

the length of the branch from A to G and is PAG.

G

C

T

T

T

A

slide66

G

C

T

T

The probability of this tree is:

Ptree=PA x PAG x PAC x PAT x PTT x PTT

Because there are 16 such scenarios, the probabilities

of each of the scenarios must be determined to

obtain the probability of the tree as follows:

Ptree = Ptree1 + Ptree2 + . . . . + Ptree 16

This is the probability for that tree for observing the

data at one site, the site marked in red.

T

A

slide67

The probability of observing all of the data at all

of the sites is the product of the probabilities for

each of the sites i from 1 to N as:

N

Ptree = Pi

i=1

slide68

Because these numbers are often too small for

most computers to handle, and because it is

computationally easier, the probability (or

likelihood) of a tree for each site i is usually

expressed as a log likelihood, lnLi, and the log

likelihood of the tree is the sum of the log

likelihoods for each of the sites as follows:

N

lnLtree = lnLi

i=1

slide69

The term lnLtree is the log likelihood of observing

the alignment under the chosen evolutionary model

given that particular tree with its branching order

and branch lengths.

ML programs seek the tree with the largest log

likelihood.

slide70

Bayesian Analysis:

Bayesian inference is based on the notion of

posterior probabilities: probabilities that are

estimated, based on some model (prior expectations),

after learning something about the data.

For example, if you are tossing coins, your model

might be that 90% are true coins and 10% are coins

that are biased to turn up heads 80% of the time.

slide71

Suppose you are blindfolded and asked to pick a coin

at random; then you are asked “What is the

probability that this coin is a biased coin?”

Having nothing more to go on than your model that

90% of the coins are true, your obvious answer is 0.1.

If, however, you are allowed to toss the coin you

chose 10 times and then are asked the probability

that it is biased, you would revise your estimate

based on your model of the expected distribution of

outcomes from true and biased coins, and your

expectations of the initial proportion of true coins.

slide72

The probability you estimate after observing the

outcomes -- the posterior probability -- should

be a better estimate than the 0.1 probability

you estimated with no knowledge.

Suppose you observe the following results of your

coin tosses: HHTHHTTHHH.

We will use X to symbolize that result.

slide73

The probability of that result given that the coin

is true -- symbolized P[X|True] where | means

“given that” is:

P[X|True] = 0.510 = 9.76 X 10-4.

The probability of that result given a biased coin is:

P[X|Biased] = 0.87 X 0.23 = 1.6 X 10-3.

The posterior probability that the coin is biased is

given by the Bayes formula as:

slide74

P[X|Biased] x P[Biased]

P[X|Biased] =

(P[X|Biased] x P[Biased]) + (P[X|True] x P[True])

1.67 x 10-3 X 0.1

P[X|Biased] =

(1.67 x 10-3 X 0.1) + (9.76 x 10-4 X 0.9)

Thus, P[Biased|X] = 0.13 and your estimate of the

probability that this is a biased coin has increased

from 0.1 to 0.13 based on your observation of

results.

slide75

Bayesian analysis of phylogenies is similar to ML

in that the used postulates a model of evolution

and the program searches for the best trees that

are consistent with both the model and with the

data (the alignment).

It differs somewhat from ML in that while ML

seeks the tree that maximizes the probability of

observing the data given that tree, Bayesian

analysis seeks the tree that maximizes the

probability of the tree given the data and the model

for evolution.

slide76

In essence, this rescales likelihood to true

probabilities in that the sum of the probabilities

over all trees is 1.0 under the Bayesian approach,

which in turn permits using ordinary probability

theory to analyze the data.

Like Parsimony and ML, the Bayesian method is

character-based and is applied to each site along

the alignment.

slide77

Gracilinanus

T. macrura

1.0

97

100

98

T. pusilla

T. pallidior

1.0

99

100

100

T. venustus

T. sponsoria

0.78

57

68

91

1.0

99

93

99

T. cinderella

T. cinderella

1.0

93

100

100

T. sponsoria

1.0

68

93

87

1.0

97

100

100

T. pusilla

T. pallidior

1.0

<50

87

99

T. tatei

1.0

63

99

100

T. elegans

0.95

55

88

100

1.0

57

97

100

1.0

91

97

100

T. pallidior

Bayesian

Maximum likelihood

unweighted parsimony

NJ-Kimura 2 parameter

slide78

Evolutionary Significant Units (ESUs)

Management Units (MUs) and

slide79

Although genetics has assumed an important role

in conservation biology, genetic surveys of

managed species are far from routine and there

is a perception that genetic analyses are of more

significance to long-term than short-term needs

and thus, are of lower priority than demographic

analysis.

Why are the theory and practice so far apart?

slide80

Moritz suggests that it is because the relevance

of genetic analyses to practical issues in

wildlife management have not been adequately

explained and demonstrated.

mtDNA is a powerful tool in evolutionary biology

because:

--rapid rate of base substitutions

--effectively haploid and maternal inheritance

reduces Ne and increases sensitivity to

genetic drift.

--ease of isolation and manipulation.

slide81

mtDNA can produce results of considerable

practicle importance, but the conservation goals

must be clearly defined first and the analyses

designed to fit the goals.

It is important to distinguish between:

Gene Conservation -- the use of genetic

information to measure and manage genetic

diversity for its own sake.

slide82

Molecular Ecology -- genetic analyses as a

complement to ecological studies of demography.

In many respects, molecular ecology is more

straight forward and is of more use to wildlife

managers faced with short-term management

priorities.

slide83

Gene Conservation: Measuring & Managing

Genetic Diversity

With few noticeable exceptions, such as

translocations, managing genetic diversity in so far

as it relates to conserving evolutionary potential, is

more relevant to long-term planning and policy

than to short-term management of threatened

populations.

slide84

mtDNA has been used in 3 ways in this context:

To measure genetic variation within populations,

especially ones thought to have declined

recently.

Identifying evolutionary divergent sets of

populations, including the resolution of

Evolutionary Significant Units.

to assess conservation value of populations or

areas from an evolutionary or phylogenetic

perspective.

slide85

Genetic Variability within Populations:

A common aim of quantifying mtDNA variation within

populations is to test for the loss of genomic

variability, perhaps as a consequence of reduction

in population size.

This will have conservation significance if the loss of

variation translates to reduced individual fitness.

slide86

This is a weak application of mtDNA because of the

lack of any theoretical or empirical evidence for a

strong correlation between mtDNA diversity and

diversity in the nuclear genome.

For example, low mtDNA diversity has been

reported in rapidly expanding species such as

northern elephant seals and parthenogenetic gekos

whereas moderate to high mtDNA diversity has

been observed in declining species subjected to

intense harvesting such as coconut crabs,

humpback whales or in species otherwise suggested

to be inbred.

slide87

Low mtDNA diversity is correlated with low

nuclear gene diversity is some case but not others.

These observations indicate that putting

management priorities on the basis of within

population mtDNA diversity is inappropriate.

mtDNA and the identification of Evolutionary

distinct populations.

A prerequisite for managing biodiversity is the

identification of populations with independent

evolutionary histories.

slide88

Such groupings are variously referred to as species,

subspecies, or evolutionary significant units (ESUs).

Following from the Rio Biodiversity Convention,

genetically divergent populations increasingly are

being recognized as appropriate units for

conservation, regardless of their taxonomic status.

mtDNA phylogenies can provide unique insights into

population history and can suggest hypotheses about

the boundaries of genetically divergent groups

(i.e., cryptic species).

slide89

However, mtDNA must be used in conjunction with

nuclear markers to identify evolutionary distinct

populations for conservation because given the

lower effective number of genes or greater

dispersal by males than females, mtDNA can

diverge while nuclear genes do not.

This is exemplified by the ring species Ensatina

eschscholtzii.

slide90

Allozyme

Group B

Allozyme Group A

oreg

xan

esch

pic

klau

plat

cro

mtDNA

= evolutinary

entities

Simplified mtDNA phylogeny from different

subspecies of the salamander ring species

E. schscholtzii overlain with major allozyme groups.

slide91

Recognition of ESUs:

The concept of an evolutionary significant unit (ESU),

a set of populations with a distinct, long-term

evolutionary history, as a focus of conservation

effort fits well with the goal of recognizing and

maintaining biodiversity.

However, the criteria for defining an ESU remains

to be established.

slide92

It has been suggested that thresholds range

from any population that “contributes substantially

to the overall genetic diversity of the species and

is reproductively isolated” to “populations showing

phylogenetic distinctiveness of alleles across

multiple loci.”

The question that plagues the approach is “How

much difference is enough?”

slide93

There is no theoretical or empirical justification

for setting an amount of sequence divergence

beyond which a set of populations is recognized as

an ESU, although comparisons to divergences within

an among related species may provide an empirical

yardstick.

slide94

One approach to defining an ESU is to consider

the geographic distribution of alleles in relationship

to their phylogeny, the rationale being that gene

flow must be restricted a long period

(2 - 4 Ne generations) to create phylogeographic

structuring of alleles.

This suggests a qualitative criterion -- ESUs should

show complete monophyly of mtDNA alleles --

thereby avoiding the quantitative question of

“How much is enough?”.

slide95

However, this criterion may be to stringent given

that well characterized species with paraphyletic

mtDNA lineages have been documented.

A less stringent criterion would be significant,

but not necessarily absolute, phylogenetic

separation of haplotypes between populations.

As already stressed, it is important to seek

corroborating evidence from nuclear loci and Avise

and Ball (1990) suggest that ESUs should exhibit

congruent phylogenetic structure with other genes.

slide96

However, alleles of nuclear genes are expected to

take substantially longer to show phylogenetic

sorting between populations or species because of

the larger effective population size and slower

neutral mutation rate.

Defining evolutionary conservation value of

populations or areas:

An extension of the use of mtDNA variation to

recognize ESUs is to explicitly define conservation

value from an evolutionary perspective.

slide97

It has been proposed that phylogenetic uniqueness

should be considered in prioritizing species for

management and this concept has been modified to

take account of evolutionary distance and is

particularly well suited to molecular data.

An exciting application of mtDNA phylogeography

is to define geographic regions within which

multiple species have genetically unique populations

or ESUs; moving from species to community

genetics.

slide98

This involves testing for congruence of

phylogeographic patterns among species to define

geographic regions within which a substantial

proportion of species have had evolutionary histories

separate from their respective conspecifics.

For example, analysis of mtDNA diversity in birds

and skinks endemic to the wet tropical rainforests

of north-eastern Australia have revealed a

geographically congruent genetic break on either

side of a dry corridor.

slide99

The significance of this for conservation is

obvious -- regions with a high proportion of ESUs

should be accorded high conservation priority

even if they do not have an array of endemic

species as recognized by conventional methods.

This discussion on conservation “value” skirts

some basic philosophical and ethical issues:

What do we mean by the “s” in ESU?

Can we justify ranking species according to a

measure of molecular divergence?

slide100

We can only measure evolutionary significance or

value in terms of past history, the proportion of a

species total genetic diversity represented by a

particular set of populations.

We cannot, however, predict which, if any, of these

units will diversify to produce future biodiversity.

slide101

Therefore, in the face of these inescapable

uncertainties, we must be very clear about the

nature of the advice we are providing when we

discuss conservation priorities from a molecular

evolutionary perspective.

slide102

Molecular Ecology:

This second general area of application uses

genetics as a tool for ecologists, in particular:

to define the appropriate geographic scale

for monitoring and managing.

2. to provide a means for identifying the origin

of individuals in migratory species.

to test for dramatic changes in population

size and connectedness.

slide103

In general, these applications are conceptually

simpler and much more relevant to short-term

management issues than are those related to gene

conservation.

Defining Management Units:

A great deal of effort is spent on monitoring

populations as part of the species recovery

process. Yet, too often, little consideration is

given to the appropriate geographic scale for

monitoring or management.

slide104

An exception is with fisheries, where it has long

been recognized that species typically consist of

multiple stocks that respond independently to

harvesting and management.

A simple but powerful and practical application of

genetics is to define such Management Units (MUs)

or stocks, the logic being that populations that

exchange so few migrants as to be genetically

distinct will also be demographically independent.

slide105

In contrast to ESUs, MUs are defined by significant

divergence in allele frequencies, regardless of the

phylogeny of the alleles because allele frequencies

will respond to population isolation more rapidly

than phylogenetic patterns.

mtDNA is especially useful for detecting boundaries

between MUs because it is usually more prone to

genetic drift than nuclear loci; meaning that a

greater proportion of the variation is distributed

between populations.

slide106

So long as variation exists, differences between

populations will be more readily detected with

mtDNA than with nuclear genes, an important

consideration when sample sizes are limited as

is often the case with threatened species.

Identification & Use of Genetic Tags:

A practical and exciting use of genetics for short-

term management is to provide a source of

naturally occurring genetic tags, genetic variants

that individually or in combination diagnose

different MUs.

slide107

Genetic tags are indelible, present in all members

of a population at all ages, and can be used to

determine the source(s) of animals in harvest,

international commerce, or areas subjected to

impacts or management.

Genetic tags are particularly useful for migratory

species where impacts in one area (e.g., feeding

ground) can affect one or more distant MUs.

slide108

Genetic tags are most effective where the

variation within areas is low relative to that

between areas, with the ideal situation being fixed

genetic differences.

Where MUs are characterized by differences in

allele frequencies of shared alleles, maximum

likelihood methods can be used to estimate the

contribution of various MUs to a sample of

individuals taken from a particular feeding ground,

migratory route, or commercial harvest.

slide109

Another problem faced by wildlife managers is

assessing the degree to which populations are

connected by migration and are changing in size.

Estimating these parameters via ecological studies

is an important but, very difficult and expensive

exercise, prompting a search for indirect methods

based on patterns of genetic variation.

At the same time, there has been rapid development

of methods for using information on allele

distributions and relationships to infer long-term

migration rates and trends in Ne.

slide110

Although these new statistical tools can provide

insight into the long-term behavior of populations,

it is not clear that they can produce information

relevant to short-term management, especially

where populations are fluctuating in size and/or

connectedness as is often the case in

conservation studies.