Ferramentas moleculares e estatsticas para estudos de estrutura de populaes

1
Ferramentas moleculares e estatísticas para
estudos de estrutura de populações

• Guillermo Ortí
• University of Nebraska
• Lincoln, NE, USA

2
PDF versions of papers and powerpoints are (will
be) downloadable fromhttp//golab.unl.edu/teachi
ng
3
Comparison of some markers
Other criteria Automation? Sequence info needed?
Start-up costs? Radioactivity needed?
4
Case studies
8 microsatellite loci Individuals (22-105 per
locality) sampled from 19 localities upstream and
downstream of dam (constructed in 1952) Test 4
hypotheses about the origin of fish congregating
at the dam site during breeding season every year
5
DATA ANALYSIS Basic population parameters, Hs
(heterozygosity), allele freq, Fst, deviations
from H-W, genetic distances among populations
(GENEPOP) Assignment tests (Paetkau et al 1997
online calculator) to determine the most probable
origin of each individual based on expected
genotypic freq of each sample. Gives proportion
of indiv from each population (locality)
assigned to every other population
6
Assignment tests are consistent with result Dam
fish are more likely to originate from upstream
populations
7
Assignment tests rely on previously defined
populations, but this definition could be
subjective or based on unreliable information
(typically geographic location, morphotype,
etc) So, how well does this initial assignment
correspond to genetic information? Some species
may have cryptic populations, not easily
defined by any visible trait(s), but have strong
genetic signatures
8

• Clustering methods typically use a distance
  measure (e.g. chord distance) among individuals
  and a clustering algorithm, UPGMA or NJ.
  These are appealing and easy to apply, good
  exploratory tools but have shortcomings
• Clusters obtained may depend on distance measure
  or clustering algorithm used
• Difficult to assess confidence (bootstrap?).
• Difficult to add other sources of information,
  such as geographical information of the samples
• Model based methods assume that observations
  from each cluster are random draws from a
  parametric model. Inference of the parameters
  corresponding to each cluster is done jointly
  with inference of cluster membership for each
  individual (ML or MCMC approach). P(X Z, P)
  where X is the genotype of an individual, Z the
  (unknown) population of origin, and P the
  (unknown) allelic freq of all populations.

9
Taita bird (Turdus helleri) in Africa Sampled
from 4 locations 7 microsatellite loci used NJ
tree of individuals Based on pairwise dist
10

• STRUCTURE
• Can accurately cluster individuals into their
  appropriate populations, even with modest number
  of loci.
• May be applied to Codominant markers such as
  microsatellites, RFLP, SNPs.
• Assumes H-W equilibrium and unlinked loci
• Accuracy depends on number of indiv sampled,
  number of loci, degree of admixture and allelic
  freq divergence among subpopulations
• Assigns individuals to subpopulations
• May identify migrants (with incorporation of
  information on geographic location) or hybrids

11
Study based on 18 microsatellite markers scored
for 380 individuals Use of clustering methods and
STRUCTURE to detect discrete genetic
populations Inference of past bottleneck effects
(demography) using three different approaches
extent and timing of demographic events estimated
(coalescent methods) Good source of analytical
tools currently available to apply to
microsatellite data
12
Comparison of some markers
Other criteria Automation? Sequence info needed?
Start-up costs? Radioactivity needed?
13
DNA sequence analysis

• Homologous sections of the genome are amplified
  (PCR) and directly sequenced for comparison.
• Easily scored, high-quality information
  sometimes alignment is a problem
• Automated procedures available (but expensive)
• Codominant marker but cloning is necessary to
  separate alleles from heterozygotes
• Universal primers facilitate sequencing for
  many taxa without prior knowledge of their DNA
  sequence (not always true)
• Once polymorphism has been identified by
  sequencing, other approaches can be used to
  screen large number of individuals (PCR-RFLP,
  SNPs, SSCP)

14
DNA sequence analysis

• APPLICATIONS
• Population diversity and structure (subdivision)
• Hybridization, introgression (among species or
  populations), gene flow
• Greatest value for phylogenetic and
  phylogeographic analysis
• ANALYSIS
• Population genetic parameters, genetic distances,
  Fst, AMOVA, etc
• Neutrality tests (Fu, Fu an Li)
• Phylogenetic inference interpretation of
  phylogeny
• Coalescent approaches becoming more common,
  increasing power of interpretation

15
mtDNA phylogeography...
16
Interspecific phylogeny

• Foundations of cladistic analysis developed by
  Hennig 1966.
• Cladogenesis only above the species level
• Species are the minimal units for phylogenetic
  analysis
• Other (many) species concepts available...

17
Difficulties of interspecific methods at the
intraspecific level

• Few characters for analysis (less variation
  within species than between species
• Ancestors not assumed to be extinct, most alleles
  exist as sets of multiple, identical copies
  because of past DNA replication
• As mutation occur to create new alleles, they
  rarely result in the extinction of the ancestral
  allele.
• Coalescent theory predicts that the most common
  allele in the gene pool will be the oldest, and
  most of these will be interior nodes of the
  haplotype tree
• Multifurcations (rather than strict bifurcations)
  are expected since common alleles will mutate
  many times

18
Big problem with gene genealogies is recombination

• Traditional methods assume that recombination
  does not occur (at all levels)
• Need to test for recombination before attempting
  phylogenetic analysis
• A methods that accounts for all these issues was
  develop to study intraspecific gene genealogies
  (TCS 1992)

19
Intraspecific gene genealogy (parsimony network)
20
Nested Clade Analysis...

• Nested clades in an mtDNA network for tiger
  salamanders collected in the central USA
  (Templeton et al 1995)
• Circles encompassing letters a-v indicate
  observed haplotypes
• Branches connect haplotypes differing by a single
  mutational step
• Differences between subspecies are 14 steps.
• Nested clades of increasing hierarchical level
  are indicated by numbers

21
Nested Clade Analysis...
Recent paper criticized strict interpretation of
results obtained with this technique
22
Nested Clade Analysis...
A reply was prompt by the lead author proposing
NCA (Alan Templeton)
23
Case study comparing mtDNA and nucDNA sequence
variation for phylogeography