Patterns of divergent selection from combined DNA barcode and phenotypic data

Tim Barraclough, Imperial College London

Goals:

1) Delimit evolutionary species

independent arenas for selection and drift

Goals:

1) Delimit evolutionary species

independent arenas for selection and drift

2) Identify the processes generating diversity

separate demography

reproductive isolation

divergent selection

DNA barcodes

DNA sequence data sampled at the individual level across an entire clade

• sample certain no. individuals per taxonomic species

• environmental samples irrespective of known species

DNA barcodes

DNA sequence data sampled at the individual level across an entire clade

• sample certain no. individuals per taxonomic species

• environmental samples irrespective of known species

new resource linking macro- and population genetic questions

DNA barcodes: limitations

• How to detect evolutionary species?

Rely on traditional species

Phenetic approaches, e.g. 2% sequence divergence.

Population models => prior guess on minimum units (c.f. bacteria)

=> multilocus, => parameter-rich

Sampling?

DNA barcodes: limitations

• Single marker

1) Species tree versus gene treemultilocus approaches

2) Arbitrary or neutral markers

No information on adaptive variationniche traits, those involved in R.I.

Goals:

1) Delimit independently evolving species

2) Identify the processes generating diversity

divergent selection

1) Delimit independently evolving species

Prediction: genetic clusters separated by longer internal branches

Conservative - missrecent

Assumptions

Just uses DNA variation

Statistical approach

Null model:Entire sample derives from single species, i.e. no independently evolving subsets of individuals

Single coalescent process

Likelihood of waiting times

L(x i) (ni(ni 1))pe (ni(ni1))p x i

1

2N e

Statistical approach

Null model:Entire sample derives from single species, i.e. no independently evolving subsets of individuals

Single coalescent process

L(x i) (ni(ni 1))pe (ni(ni1))p x i

1

2N e

scaling parameter

p<1 excess of old branching events, p>1 excess of recent

Statistical approach

Alternative model: separate independently evolving species

Within species branches coalescence

Between species speciation, extinction

Between species branchingWithin species branching

Alternative model: separate populations

L(xi )bebxi

b* k1

(ni,k1

)pk1 ( j (n i,j (n i,j 1))

pj )

j1, k

1. Label which branches are within v. between species

2. Set of independent coalescent processes in each species

3. Generalized Yule model for between species branchingp=1 constant speciation rate modelp>1 increasing speciation rate, background extinctionp<1 slowdown, incomplete sample of species

(Mixed Coalescent Yule model, Pons et al. 2006. Syst Biol. 55:559-609)

Alternative model: separate populations

Implementation

Optimize which nodes define separate species, e.g. sliding threshold or more complex

Confidence intervals on delimitation

Hypothesis testing

Alternative model: separate populations

Example

Tiger beetles fromAustralian salt lakes

468 individuals

48 +2/-4 genetic clusters

Fitted parameters:Growing populationsor selective sweeps

Pons et al. 2006. Syst Biol. 55:559-609)

Alternative model: separate populations

Limitations

Current implementation uses sliding threshold

Identical individuals

Sampling

Not exact (but generalized)

Conservative, but could focus e.g. multi-locus

Correcting for mtDNA rate variation

Goals:

1) Delimit independently evolving species

2) Identify the processes generating diversity

divergent selection

Focus on a trait or traitsecomorphologyreproductive morphologybehaviour, defensive chemicals

etc.

Divergent selection

Character under divergent selection displays greater ratio of

inter-group variationinter-group variation

Than neutrally evolving characters

Can compare variation in morphological traits to variation of arbitrary DNA markers

Qst-Fst Fontaneto et al. 2007. PLoS Biology 5:e87

Divergent selection

Compared to clusters identified from mtDNA

1) Coincident with clusters

2) Acting at broader level (uniform selection across entire clade or sub-clades)

3) Acting within clusters - recently formed species or adaptive polymorphism

Divergent selection

Example: divergent selection on feeding morphology in bdelloid rotifers

Fontaneto et al. 2007. PLoS Biology 5:e87

Significant pattern of clustering

Significant pattern of clustering

ML solution: 13 clustersCOI Pairwise within = 1.5%COI Pairwise between = 16%

H0:H1, p<0.0001

Some traditional species contain several clusters

TraditionalSpecies are morphologicalclusters

Mapped rate of evolution of trophi size and shape relative to silent mtDNA change

Null model: one rate across DNA tree

Alternatives: 1) Within versus between species2) Within versus between clusters3) Three rates

Schluter et al. 1997.

Results

Significant evidence for divergent selection on trophi size and shape between taxonomic species, not clusters.

Sexual organisms?

Assumptions

Assume additive genetic variation environmental variation might inflate intra

Limited to measurable traits, measurement error will inflate intra

Prospects?

DNA barcode data as framework to explore selection on morphological traits of voucher specimens