Transcript Gallwasp Barcoding

BANBURY 3
The Barcode Gap
Speciation or Phylogeography?
?
Graham Stone, Richard Challis, James Nicholls, Jenna Mann, Sonja Preuss
Mark Blaxter
Institute of Evolutionary Biology, Edinburgh University
BANBURY 3
Phylogeography and DNA barcoding often use the same
tools, but have different aims.
Barcoding
• aims to identify species-specific sequences at a single
locus.
• would like to capture the full diversity of sequence
variation inherent in a taxon, but usually does not.
• works best when ancestral polymorphism between sister
lineages has been completely sorted, creating
monophyletic sister clades and a ‘barcoding gap’.
BANBURY 3
Mitochondrial barcoding traumas
 Incomplete sorting of ancestral polymorphism.
 Barcodes jump between species.
 Introgression and sorting may only become obvious if
you sample closely related taxa in depth. You cannot
know the scale of this problem by sampling one taxon.
 Empiricially, barcoding generally works.
BANBURY 3
Phylogeography
•Reconstructs the spatial relationships between
lineages over time
•Requires extensive within-species sampling
•Commonly combines mitochondrial and nuclear markers
(allozymes, microsatellites, sequence)
•Struggles to find nuclear loci polymorphic enough to
allow direct comparison with mitochondrial
sequence data.
BANBURY 3
Phylogeographic analyses of closely related species can,
coincidentally, allow stringent testing of the of the DNA
barcoding approach.
BANBURY 3
Oak gallwasps
• 1000 species worldwide, highly hostplant specific
• induce characteristic gall structures
• Nuclear gallwasp genes determine gall structure: distinctive
gall morphologies reliably identify species.
They support ecologically closed communities of natural enemies
BANBURY 3
Individual species are found across the Western Palaearctic
BANBURY 3
..with populations in multiple glacial refugia
Turkish refugia
Iranian refugia
BANBURY 3
Comparative phylogeography
Concordance: same origin,
direction and timescale
Concordance: same origin
and direction, different
timescales/demographies
Discordance: different origin
and different direction (±
same timescale)
BANBURY 3
Andricus kollari
Gallwasp phylogeography
Allozyme allele frequency data
13 loci, 2100 individuals 70 sites
Intraspecific variation in
widespread species
corresponds to refugia
BANBURY 3
Andricus kollari
Gallwasp phylogeography
433 bp Cytb, 160 individuals
2-3MY
BANBURY 3
Andricus kollari
Introgression and backcrossing
creates barcode mismatches
433 bp Cytb, 160 individuals
2-3MY
BANBURY 3
BANBURY 3
Parasitoid phylogeography
Eastern haplotypes
350
300
Frequency
250
200
Observed
Simulated
150
100
50
0
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
1.8 MY
Number of mismatches
European haplotypes
7000
6000
Frequency
5000
4000
Observed
Simulated
3000
2000
1000
0
0
1
2
3
4
5
6
7
8
Number of mismatches
9
10
11
12
BANBURY 3
The Andricus quercuscalicis clade
..is a group of closely-related oak gallwasp species with
highly diagnostic gall phenotypes
Andricus
quercuscalicis
Andricus
quercustozae
Andricus
dentimitratus
Andricus
caputmedusae
Cytochrome b sequence resolves each species into
glacial refuge clades
Outgroup individuals from Turkey
Andricus
quercustozae
433 bp Cytochrome b
(all seqs ORF)
K2P NJ
BANBURY 3
But the geographic origin of outgroups influences
relationships between these clades
Outgroup individuals from Turkey
Outgroup individuals from C. Europe
BANBURY 3
Analyse data for all 4 closelyrelated species (n=600
individuals, 221 haplotypes)
K2P NJ
BANBURY 3
Analyse data for all 4 closelyrelated species (n=600
individuals, 221 haplotypes)
Sequence divergence within
this whole group matches that
within single gallwasp species
(e.g. A. kollari)
K2P NJ
BANBURY 3
Molecular Operational Taxonomic
Units (MOTU)s identified using
MOTU_define
BANBURY 3
Identify MOTU’s using
MOTU_define
Little sign of a genuine
barcoding gap in these
data
Colour blocks
show 8bp
MOTU’s
BANBURY 3
MOTU’s rarely
correspond to species
Andricus quercustozae
Colour blocks
Colour
show
8bpblocks
show 8bp
MOTU’s
MOTU’s
BANBURY 3
MOTU’s rarely
correspond to species
Andricus caputmedusae
Colour blocks
Colour
show
8bpblocks
show 8bp
MOTU’s
MOTU’s
BANBURY 3
MOTU’s rarely
correspond to species
Andricus dentimitratus
Colour blocks
Colour
show
8bpblocks
show 8bp
MOTU’s
MOTU’s
BANBURY 3
MOTU’s rarely
correspond to species
Andricus quercuscalicis
Colour blocks
Colour
show
8bpblocks
show 8bp
MOTU’s
MOTU’s
BANBURY 3
This is true irrespective of the
threshold sequence difference for
MOTU’s
Colour blocks
Colour
show
8bpblocks
show 8bp
MOTU’s
MOTU’s
BANBURY 3
This is true irrespective of the
threshold sequence difference for
MOTU’s
Colour blocks
Colour
show
8bpblocks
show 8bp
MOTU’s
MOTU’s
BANBURY 3
.. And is true when phylogeny
reconstruction uses more complex
models
MrBAYES
HKY+G, partitioned by codon
position, parameter estimates
unlinked across partitions
Ln Bayes Factors against species
monophyly:
Andricus caputmedusae
Andricus dentimitratus
Andricus quercustozae
270
332
158
BANBURY 3
Instead, multispecies
MOTU’s correspond to
geographic regions
BANBURY 3
What generates the observed pattern?
Geographic grouping rules out sorting of ancestral
polymorphism
No evidence for role of symbionts Wolbachia,
Cardinium, Spiroplasma, Flavobacteria.
BANBURY 3
What generates the observed pattern?
Nuclear sequence data do
not support multispecies
clades sorted by refuge
702 bp Nuclear
Long wavelength
opsin gene
BANBURY 3
What generates the observed pattern?
Nuclear sequence data do
not support multispecies
clades sorted by refuge
MtDNA data
0.99
0.95
702 bp Nuclear
Long wavelength
opsin gene
BANBURY 3
What generates the observed pattern?
Nuclear sequence data do
not support multispecies
clades sorted by refuge
MtDNA data
0.99
0.95
702 bp Nuclear
Long wavelength
opsin gene
Observed mtDNA patterns
are compatible with
hybridisation and backcrossing to parental types
within refugia
BANBURY 3
Conclusions 1
How common is this pattern?
Don’t know: we need studies of appropriate taxa.
But gallwasps are unlikely to be an isolated case.
•Many radiations of species have occurred over the same
timescale, especially insects.
•Most widespread taxa show differentiation between
regional refugia.
•Little is known for most taxa about current or past
hybridisation within refugia.
•Expect to find more examples as phylogenetic density of
barcode sampling increases.
Not safe to assume it is rare.
BANBURY 3
Conclusions 2
A worst case scenario:
IF related taxa often hybridise in refugia, mtDNA
barcoding will generate a molecular taxonomy of
refugia, not species.
Barcoding gaps would then indicate
phylogeographic breaks, not speciation.
Cool for reconstructing community history, bad for
species barcodes.
BANBURY 3
Conclusions 3
1. We should check how often this occurs, even in
groups where we think barcoding works, by
sampling over the full range of species.
2. We need nuclear sequence markers to
corroborate mitochondrial barcodes.
BANBURY 3
Look on the bright side…
What could CBOL do with broader mt+nuclear
sampling that it can’t do now?
•Reduce uncertainty in identification
•Facilitate array-based barcoding
•Identify sources and demographies of key target
species.
•Assess multispecies concordance: biodiversity
hotspots.
•Enhance phylogenetic utility
BANBURY 3