Haplowebs as a graphical tool for delimiting species: a revival of

RESEARCH ARTICLE Open Access

Haplowebs as a graphical tool for delimitingspecies a revival of Doylersquos ldquofield forrecombinationrdquo approach and its application tothe coral genus Pocillopora in ClippertonJean-Franccedilois Flot1234 Arnaud Couloux2 Simon Tillier3

Abstract

Background Usual methods for inferring species boundaries from molecular sequence data rely either on genetrees or on population genetic analyses Another way of delimiting species based on a view of species as ldquofieldsfor recombinationrdquo (FFRs) characterized by mutual allelic exclusivity was suggested in 1995 by Doyle Here wepropose to use haplowebs (haplotype networks with additional connections between haplotypes found co-occurring in heterozygous individuals) to visualize and delineate single-locus FFRs (sl-FFRs) Furthermore weintroduce a method to quantify the reliability of putative species boundaries according to the number ofindependent markers that support them and illustrate this approach with a case study of taxonomically difficultcorals of the genus Pocillopora collected around Clipperton Island (far eastern Pacific)

Results One haploweb built from intron sequences of the ATP synthase β subunit gene revealed the presence oftwo sl-FFRs among our 74 coral samples whereas a second one built from ITS sequences turned out to becomposed of four sl-FFRs As a third independent marker we performed a combined analysis of two regions of themitochondrial genome since haplowebs are not suited to analyze non-recombining markers individuals weresorted into four haplogroups according to their mitochondrial sequences Among all possible bipartitions of ourset of samples thirteen were supported by at least one molecular dataset none by two and only one by all threedatasets this congruent pattern obtained from independent nuclear and mitochondrial markers indicates that twospecies of Pocillopora are present in Clipperton

Conclusions Our approach builds on Doylersquos method and extends it by introducing an intuitive user-friendlygraphical representation and by proposing a conceptual framework to analyze and quantify the congruence betweensl-FFRs obtained from several independent markers Like delineation methods based on population-level statisticalapproaches our method can distinguish closely-related species that have not yet reached reciprocal monophyly atmost or all of their loci like tree-based approaches it can yield meaningful conclusions using a number of independentmarkers as low as three Future efforts will aim to develop programs that speed up the construction of haplowebs fromFASTA sequence alignments and help perform the congruence analysis outlined in this article

BackgroundSpecies delimitation is an old issue in biology that con-tinues to attract considerable attention [1-3] as the pre-sent global biodiversity crisis makes it of paramountimportance to delineate and identify as objectively as

possible species-level taxa [4] The widespread occur-rence of cryptic species [5] and the problems they posein ecological physiological and genetic studies [6] alsocall for the establishment of methods that can beapplied by scientists from a variety of backgrounds andnot solely by specialized taxonomistsAs pointed out by de Queiroz [7] diverging lineages

acquire progressively through time a number of differentproperties that allow their recognition as distinct

Correspondence jflotuni-goettingende1Courant Research Center ldquoGeobiologyrdquo University of GoumlttingenGoldschmidtstr 3 37077 Goumlttingen GermanyFull list of author information is available at the end of the article

Flot et al BMC Evolutionary Biology 2010 10372httpwwwbiomedcentralcom1471-214810372

copy 2010 Flot et al licensee BioMed Central Ltd This is an Open Access article distributed under the terms of the Creative CommonsAttribution License (httpcreativecommonsorglicensesby20) which permits unrestricted use distribution and reproduction inany medium provided the original work is properly cited

species Among the different categories of characters(eg morphological immunological ecological or mole-cular) suitable for assessing the divergence betweenlineages DNA sequences have gained increasing popu-larity among taxonomists in recent years as they are inmany cases not influenced by environmental conditionsnor by the life stage of the organisms under scrutinymoreover gathering information on DNA markersrequires less taxon-specific training and expertise thanfor other categories of characters [8] Methods for deli-miting species based on molecular sequence data can bebroadly classified in two categories tree-based andnon-tree-based [1] Tree-based methods use modelsof sequence evolution to reconstruct nodes that repre-sent hierarchical kinship relationships among organismsin a historical perspective whereas non-tree-basedapproaches use population genetic models to look forevidence of barriers or restrictions to gene flow amongand within extant populationsEven though ldquoa highly corroborated hypothesis of

lineage separation (ie of the existence of separatespecies) requires multiple lines of evidencerdquo [9] as afirst approach it may be desirable to choose a speciesdelimitation criterion that is as general as possible andcan be used systematically Since a recent survey found23 of species reported in the literature to be eitherpoly- or paraphyletic at the various loci investigated[10] a sensitive delimitation method should be capableof detecting closely related species at an early stage oflineage divergence when most of their genetic locihave not yet reached reciprocal monophyly (Figure 1)Moreover it would be advantageous to use as a gen-eral yardstick for delimiting species a method thatdoes not make restrictive assumptions regarding thegenetics of the marker used (eg the absence of copy-number variations) and of the populations studied(eg random mating)One such method applicable to sexually reproducing

organisms was proposed 15 years ago by Doyle [11]briefly this approach uses information on the co-occurrence of alleles in the diploid phase to delineatefor each marker investigated groups of individuals shar-ing a common ldquoallele poolrdquo In Doylersquos terminologythese groups of individuals are named ldquofields for recom-binationrdquo (FFRs) following an earlier proposal by Carson[12] to consider species as groups of individuals whosealleles recombine through segregation and meiosisDoylersquos method relies on the expectation that it takesless time for diverging populations to reach mutualallelic exclusivity than reciprocal allelic monophyly(Figure 1) hence groups of individuals that do not haveany allele in common can be assumed to belong todistinct species even though they are not reciprocallymonophyletic

In Doylersquos original proposal a first step is to delineatefor each genetic locus under scrutiny single-locus FFRs(sl-FFRs) However ldquowith sufficiently fine resolution theallele pool may not extend beyond the single heterozy-gous individuals in which two alleles coexist in suchsituation ldquomany more allele pools are recognized andconcomitantly there are more FFRs each consisting of asingle individualrdquo [11] To overcome this problem Doyleproposes to use information on the co-occurrence ofalleles from different loci to lump sl-FFRs into multilo-cus FFRs (ml-FFRs) if individuals A and B belong to thesame sl-FFR for marker 1 and individuals B and Cbelong to the same sl-FFR for marker 2 then individualsA B and C belong to the same ml-FFR One drawbackof this approach however is that it combines informa-tion from all available markers which makes it difficultto judge the reliability of the ml-FFRs obtained (forinstance the inclusion of a single ancestrally shared orrecently introgressed sequence in a dataset would causethe whole analysis to yield incorrect conclusions) More-over Doylersquos method requires determining the alleles ofmany individuals for several codominant nuclear mar-kers until recently this could only be done for low-resolution andor homoplasy-plagued markers such asallozymes or microsatellites which probably explainswhy earlier attempts to use this method for species deli-mitation were unsuccessful [13-15]In the last few years new techniques have emerged

that make it possible to obtain information-rich allelicsequences from heterozygous individuals without clon-ing [1617] Sequences obtained from exon-primedintrons-crossing (EPIC) nuclear markers [18] as a resultappear now well suited for delimiting species usingDoylersquos method And since such sequences are of finite(and usually moderate) length a simple strategy toobtain sl-FFRs that accurately delineate reproductivelyisolated populations is to sequence a large number ofindividuals One may then assess the reliability of theresulting species boundaries by checking whether theyare supported by several independent molecular mar-kers a congruence approach [1920] that presents theadvantage of not mixing together information from dif-ferent sourcesHere we propose to revive and invigorate Doylersquos

approach by extending it in two directions First wepresent a reliable graphical method to delineate sl-FFRsin large datasets using haplowebs (a contraction of ldquohap-lotype websrdquo) these two-dimensional representations areobtained from conventional haplotype networks (hapl-onetsrdquo) by adding connections between haplotypesfound co-occurring in heterozygous individuals ie hap-lotypes that belong to the same allele pool sensu DoyleAnd second we introduce a procedure to complementthe ml-FFR approach (used by Doyle to infer species


Page 2 of 14

10 20 30 40 50 60

11 21 31 41 51 61

12 22 32 42 52 62

allele of species β allele of species γ

allele of species α

speciation (αrarr β + γ)

allele mutation allele extinction

T2 two species (reciprocally

monophyletic)

T1 two species (not reciprocally monophyletic)

speciation

T0 one species

Figure 1 Mutual exclusivity vs reciprocal monophyly To illustrate the concepts of mutual exclusivity and reciprocal monophyly let usvisualize how the alleles in a gene tree are distributed at the various stages of the process of speciation Unless the genetic polymorphism ofthe ancestor species is very low (for instance following a strong bottleneck event) sequencing at T0 any variable marker from a number ofindividuals of this species would yield a diversity of sequences (haplotypes 10 to 60 in pink) Following speciation the two resulting sisterspecies inherit the polymorphism of their common ancestor and are thus initially indistinguishable but their sequences immediately start todiverge as some haplotype lineages get extinct through genetic drift (lineage sorting) while others accumulate mutations If effective populationsizes are large genetic drift acts more slowly than mutations in such case the two sister species become genetically distinguishable at T1 whentheir sets of sequences are mutually exclusive (haplotypes 11 31 41 51 in red haplotypes 21 and 61 in blue) ie the two species do not shareany sequence reciprocal monophyly is reached at a later stage (T2 haplotypes 12 to 62) or may theoretically never be reached if the effectivepopulation size is infinite In all cases mutual exclusivity is reached before or at the same time as reciprocal monophyly hence mutualexclusivity is a more powerful and sensitive criterion than reciprocal monophyly to delineate species


Page 3 of 14

boundaries from several markers) with an analysis of thecongruence between markers by scoring each possiblebipartition of the set of individuals according to thenumber of independent marker supporting it These twoapproaches are illustrated with a detailed example oftheir application to taxonomically difficult corals of thegenus Pocillopora (Figure 2) collected around Clipper-ton an atoll located in the far eastern Pacific Ocean

ResultsNuclear and mitochondrial markers yield putative species-level groupings of individualsTwo nuclear markers and two fragments of the mitochon-drial genome were successfully sequenced from each ofthe 74 Pocillopora samples that we had collected aroundClipperton Haplonets were constructed for each nuclearmarker and for the concatenation of the two mitochon-drial markers the haplonets obtained from each nuclearmarker were subsequently converted into haplowebs bydrawing additional connections between haplotypes foundco-occurring in heterozygous individualsThe internal transcribed spacers (ITS) are variable

non-coding regions located in the nuclear ribosomalDNA that have been proposed as a universal species-level marker in corals [21] In the present study wefocused on the ITS2 region located between the58S and 28S ribosomal genes Fifteen different ITS2sequences were detected (Figure 3a) the most commonof which was found in 47 individuals (out of 74)whereas eight sequences were singletons (ie were onlypresent in one individual each) Visual inspection of theresulting haploweb (Figure 3b) revealed four allele pools(sensu Doyle) comprising 10 2 2 and 1 sequences thecorresponding four sl-FFRs comprised 55 17 1 and 1individuals respectivelyAs a second supposedly independent nuclear marker

we sequenced an intron of the ATP synthase b subunit(ATPSb) gene 23 distinct ATPSb haplotypes weredetected (Figure 4) the most common of which wasfound in 32 individuals whereas 9 haplotypes were sin-gletons The corresponding haploweb revealed two sl-FFRs the larger allele pool included 20 haplotypesfound in 57 individuals whereas the smaller oneincluded 3 haplotypes found in 17 individualsAs for the two mitochondrial fragments (the putative

control region and a highly variable ORF of unknownfunction [22]) haplowebs were not suited to analyzethem since they were not recombining (a single haplo-type was found in each individual as is usually the casewith mitochondrial markers) hence individuals weresimply sorted into haplogroups according to the mito-chondrial haplotype they possessed Only four differenthaplotypes were detected among the 74 samples ana-lyzed (Figure 5) the most common haplotype found in

52 coral colonies was separated by only one mutationfrom the two less common ones (respectively found in 2and 3 individuals) whereas the fourth haplotype presentin 17 individuals was 16 mutations away

The congruence between independent groupings can bequantified using bipartition scoresEach of our three independent datasets (ITS ATPSband the combined mitochondrial regions) supportedseveral possible species boundaries but combining theresults of the two nuclear markers following Doylersquosapproach yielded two ml-FFRs comprising respectively57 and 17 individuals To assess the reliability of thisresult we considered all possible bipartitions of our 74samples and scored them according to the number ofmarkers supporting them (Table 1) the mitochondrialregions were included in this analysis as a single markerindependent from the two nuclear ones Generallyspeaking the number of possible bipartitions of a set ofn objects is 2n-1-1 (if one excludes the trivial case ldquoallobjects vs none of themrdquo) hence a dataset comprisedof two sl-FFRs (or two haplogroups) supports a singlebipartition a dataset comprised of three sl-FFRs (orthree haplogroups) supports 3 bipartitions a datasetcomprised of four sl-FFRs (or four haplogroups) sup-ports 7 bipartitions etcThe ATPSb haploweb detected a single putative spe-

cies boundary in our dataset partitioning it in two sl-FFRs comprising respectively 57 and 17 individualsSince the ITS2 haploweb was comprised of 4 sl-FFRs itsupported 7 possible bipartitions of our set of 74 indivi-duals only one among which was also supported by theother nuclear marker As for the combined mitochon-drial regions they delineated four haplogroups amongour samples (as defined by the mitochondrial haplotypethey harbored) out of the 7 possible bipartitions sup-ported by this marker one was also supported by theITS2 and ATPSb datasetsAmong the 13 bipartitions supported by at least one

marker none was supported by two markers and one wassupported by all of them (Table 1) This single well-supported bipartition divides our set of samples into twopopulations of 57 and 17 individuals (Figure 6) that do notshare any allele at the three independent genetic lociinvestigated according to the mutual allelic exclusivity cri-terion (Figure 1) these two populations thus represent dis-tinct species here designated Pocillopora sp A andPocillopora sp B pending further taxonomic examination

DiscussionUsing haplowebs to delineate sl-FFRs delimiting specieswithout a priori hypothesesAs illustrated with our Pocillopora example haplowebsprovide a simple intuitive way to delineate sl-FFRs


Page 4 of 14

Figure 2 Morphological diversity of the coral genus Pocillopora in Clipperton Corals of the genus Pocillopora are common on nearly alltropical reefs except in the Caribbean Their taxonomy is extremely confusing due to their extensive phenotypic plasticity and clear-cutdiagnostic morphological characters are missing for most currently defined species [35] hence individuals that cannot be reliably identified arecommon Clipperton Island is a very good and somewhat extreme example of this confusion with different taxonomic experts havingrecognized successively three [50] one or two [51] three [52] and six [53] species among specimens collected around this atoll We illustratehere a small sample of this morphological variation from left to right and top to bottom colony 05Clip026 colonies 05Clip052 (left side of thepicture green) and 05Clip053 (right side of the picture brown) colony 05Clip045 colony 05Clip018 colony 05Clip019 colony 05Clip002According to the results of the molecular analyses presented in this article the colonies in the four first photographs belong to one species(Pocillopora sp A) and the last two colonies belong to another (Pocillopora sp B) a delineation far from obvious based on their morphology


Page 5 of 14

361

603

165

166

97

352

353

224

150

170

605

359

135 316

470

390

470

390

119

265

175

390

470

359

359

05Clip001 05Clip006 05Clip013 05Clip014 05Clip015 05Clip018 05Clip026 05Clip027 05Clip028 05Clip029 05Clip030 05Clip031 05Clip032 05Clip033 05Clip035 05Clip036 05Clip037 05Clip038 05Clip039 05Clip040 05Clip041 05Clip042 05Clip043 05Clip044 05Clip045 05Clip047 05Clip048 05Clip049 05Clip050 05Clip051 05Clip052 05Clip053 05Clip055 05Clip057 05Clip058 05Clip059 05Clip060 05Clip062 05Clip063 05Clip064 05Clip081 05Clip085 05Clip086 05Clip089 05Clip090 05Clip092 05Clip093 05Clip094 05Clip095 05Clip097 05Clip098 05Clip099 05Clip100 05Clip101 05Clip102

05Clip002 05Clip003 05Clip005 05Clip007 05Clip012 05Clip019 05Clip021 05Clip022 05Clip034 05Clip046 05Clip061 05Clip079 05Clip080 05Clip082 05Clip087 05Clip096

05Clip103

05Clip016

05Clip056

361

603

165

166

97

352

353

224

150

170

605

359

135 316

470

390

470

390

119

265

175

390

470

359

359

(a)

(b)Figure 3 ITS haplonet and haploweb (a) As a first step in the analysis a haplotype network (in short ldquohaplonetrdquo) was build from thealignment of all ITS sequences Since 31 individuals were heterozygous and 43 were homozygous for this marker the alignment comprised 105sequences among which 15 different haplotypes could be distinguished (represented as circles on the haplonet with diameters proportional tothe number of individuals harboring each of them) Each haplotype is connected to one or several others by straight lines representing theevolutionary paths inferred by the network-building algorithm (numbers in red on the lines represent mutated positions in the alignment) (b) Asa second step the haplonet was converted into a haplotype web (in short ldquohaplowebrdquo) by adding curves connecting haplotypes found co-occurring in heterozygous individuals (the width of each curve is drawn proportional to the number of heterozygotes harboring the twohaplotypes it connects) This allowed us to delineate 4 pools of co-occurring haplotypes (enclosed in green dashes on the figure) To each ofthese 4 allele pools corresponds a group of individuals (called single-locus fields for recombination or sl-FFR in Doylersquos terminology) whosenames are listed inside boxes with arrows pointing on the corresponding allele pool these 4 sl-FFRs comprised respectively 55 17 1 and 1individuals


Page 6 of 14

394 395 462

147 148

207 216

260 323

40 131

204 212

228 291

357 380

385 396

417 460

473

154

79 373 462

477

40

177

456

226

373 449

142 150

279 401

487

376 449 285

449 376 40

396

1 106

122 123

200

154

178 224 319 385 455

125 370 435

145 374

40 376

129 208

207 301

129 224 279 280 342

234 280

385

455 473

475

483

120 120 120

172 373

394 395 462

147 148

207 216

260 323

40 131

204 212

228 291

357 380

385 396

417 460

473

154

79 373 462

477

40

177

456

226

373 449

142 150

279 401

487

376 449 285

449 376 40

396

1 106

122 123

200

154

178 224 319 385 455

125 370 435

145 374

40 376

129 208

207 301

129 224 279 280 342

234 280

385

455 473

475

483

120 120 120

172 373

05Clip001 05Clip006 05Clip013 05Clip014 05Clip015 05Clip016 05Clip018 05Clip026 05Clip027 05Clip028 05Clip029 05Clip030 05Clip031 05Clip032 05Clip033 05Clip035 05Clip036 05Clip037 05Clip038 05Clip039 05Clip040 05Clip041 05Clip042 05Clip043 05Clip044 05Clip045 05Clip047 05Clip048 05Clip049 05Clip050 05Clip051 05Clip052 05Clip053 05Clip055 05Clip056 05Clip057 05Clip058 05Clip059 05Clip060 05Clip062 05Clip063 05Clip064 05Clip081 05Clip085 05Clip086 05Clip089 05Clip090 05Clip092 05Clip093 05Clip094 05Clip095 05Clip097 05Clip098 05Clip099 05Clip100 05Clip101

05Clip102


05Clip103

(a)

(b)

Figure 4 ATPSb haplonet and haploweb (a) For the ATPSβ marker there were 28 homozygotes and 46 heterozygotes hence the alignmentcomprised 120 sequences among which 25 distinct haplotypes could be distinguished (b) There were 2 pools of co-occurring haplotypes(enclosed in green dashes on the figure) in the ATPSβ haploweb corresponding to 2 sl-FFRs that comprised respectively 17 and 57 individuals


Page 7 of 14

from a set of nuclear sequences and represent themgraphically In this article we chose to represent haplo-webs by adding connections between co-occurringhaplotypes on top of haplotype networks (since net-work-building algorithms are arguably more suited toreconstruct intra-specific genealogies than are phyloge-netic methods [23]) but it should be emphasized that

haplowebs can also be drawn atop phylogenetic trees ifone wishes to do soUsual tree-based methods can only delineate species that

are reciprocally monophyletic (except for some species-tree approaches based on the coalescent that show greatpromise for the future [24] but still often return incorrectresults when applied to species delimitation [25]) whereas

119

148

191

205

218

235

266

271

314

322

345

356

363

629

655

1810

1455

1447

05Clip002 05Clip003 05Clip005 05Clip007 05Clip012 05Clip019 05Clip021 05Clip022 05Clip034 05Clip046 05Clip061 05Clip079 05Clip080 05Clip082 05Clip087 05Clip096 05Clip103

05Clip001 05Clip006 05Clip013 05Clip014 05Clip015 05Clip016 05Clip018 05Clip026 05Clip028 05Clip029 05Clip030 05Clip031 05Clip032 05Clip033 05Clip035 05Clip036 05Clip037 05Clip038 05Clip039 05Clip040 05Clip041 05Clip042 05Clip043 05Clip044 05Clip045 05Clip047 05Clip049 05Clip050 05Clip052 05Clip053 05Clip055 05Clip057 05Clip058 05Clip059 05Clip060 05Clip062 05Clip063 05Clip064 05Clip081 05Clip085 05Clip086 05Clip089 05Clip090 05Clip092 05Clip093 05Clip094 05Clip095 05Clip097 05Clip098 05Clip099 05Clip101 05Clip102

05Clip048 05Clip056

05Clip027 05Clip051 05Clip100

Figure 5 Mitochondrial haplonet Since there was only one mitochondrial sequence per individual the haplonet obtained from this markercould not be converted into a haploweb (there were no co-occurring haplotypes) However 4 haplogroups could be distinguished thatcomprised respectively 52 17 3 and 2 individuals

Table 1 Bipartitions of our set of samples and the molecular markers supporting them

Bipartitions ITS2 ATPSb ORF+CR support

1 G vs all other samples radic radic radic 100

2 05Clip016 vs all others radic 33


4 05Clip016+05Clip056 vs all others radic 33

5 G+05Clip016 vs all others radic 33


7 G+05Clip016+05Clip 056 vs all others radic 33


9 05Clip027+05Clip051+05Clip100 vs all others radic 33

10 05Clip048+05Clip027+05Clip056+05Clip051+05Clip100 vs all others radic 33

11 G+05Clip048+05Clip056 vs all others radic 33

12 G+05Clip027+05Clip051+05Clip100 vs all others radic 33

13 G+05Clip048+05Clip027+05Clip056+05Clip051+05Clip100 vs all others radic 33

G = 05Clip002+05Clip003+05Clip005+05Clip007+05Clip012+05Clip019+05Clip021+05Clip022+05Clip034 +05Clip046+05Clip061+05Clip079+05Clip080+05Clip082+05Clip087+05Clip096+05Clip103


Page 8 of 14

05Clip00205Clip003 05Clip005

05Clip007 05Clip012 05Clip019 05Clip021 05Clip022 05Clip034 05Clip046 05Clip061 05Clip079 05Clip080 05Clip082 05Clip087

05Clip096 05Clip103


05Clip014 05Clip015 05Clip01805Clip026 05Clip028 05Clip029 05Clip03005Clip031 05Clip032 05Clip033 05Clip03505Clip036 05Clip037 05Clip038 05Clip03905Clip040 05Clip041 05Clip042 05Clip04305Clip044 05Clip045 05Clip047 05Clip04905Clip050 05Clip052 05Clip053 05Clip05505Clip057 05Clip058 05Clip059 05Clip06005Clip062 05Clip063 05Clip064 05Clip08105Clip085 05Clip086 05Clip089 05Clip09005Clip092 05Clip093 05Clip094 05Clip095

05Clip097 05Clip098 05Clip09905Clip101 05Clip102

05Clip048


05Clip016

05Clip056

1

2

3

4

5

6

7

8

9

10

11

1213

05Clip00205Clip003 05Clip005 05Clip007 05Clip012 05Clip019 05Clip021

05Clip022 05Clip034 05Clip046 05Clip061 05Clip07905Clip080 05Clip082 05Clip087 05Clip096

05Clip103


05Clip014 05Clip015 05Clip016 05Clip018 05Clip026 05 Clip027 05Clip028 05Clip029 05Clip030 05Clip031 05Clip032 05Clip033 05Clip035 05Clip036 05Clip037 05Clip038 05Clip039 05Clip040 05Clip041 05Clip042 05Clip043 05Clip044 05Clip045 05Clip047 05Clip048 05Clip049 05Clip050 05Clip051 05Clip052 05Clip053 05Clip055 05Clip056 05Clip057 05Clip058 05Clip059 05Clip060 05Clip062 05Clip063 05Clip064 05Clip081 05Clip085 05Clip086 05Clip089 05Clip090 05Clip092 05Clip093 05Clip094

05Clip095 05Clip097 05Clip098 05Clip099 05Clip100 05Clip101 05Clip102

1

(a)

(b)

Pocillopora sp A

Pocillopora sp BFigure 6 Bipartition reconciliation (a) In order to determine graphically the number of well-supported groups of individuals bipartitions canbe represented on a two-dimensional figure as lines or curves (each of them dividing the group of samples analyzed into two) Some groups ofindividuals observable in such graph do not correspond to sl-FFRs but rather to intersections of sl-FFRs obtained from different markers herethis is the case of individual 05Clip048 and of the large group of samples in the upper right corner of the figure whose support is equal to zeroas these groups are not supported by any single marker Each bipartition is numbered from 1 to 13 to facilitate comparison with the respectivelines of Table 1 and is drawn with its thickness proportional to the number of independent datasets supporting it (b) Bipartitions below anarbitrary support threshold (here 50) may be omitted from such graph for the sake of clarity


Page 9 of 14

most non-tree-based methods are statistical in nature andrequire large sample sizes and numbers of markers inorder for meaningful conclusions to be reached Anotherissue with such statistical approaches is that genotypeshave to be determined for each individual sampled whichposes problems in cases of polyploiumldy aneuploidy copy-number variation or chimerism when individuals comprisevariable numbers of haplotypes and the genotypes cannotbe determined (as we experienced in a previous study ofPocillopora corals from Hawaii [26]) In contrast the deli-neation of allele pools using Doylersquos method is based onlyon haplotype presenceabsence information and thus doesnot require knowledge of the respective amount of eachhaplotype in the genotypeEven though haplowebs have their most obvious appli-

cation in the analysis of nuclear sequence markers onemay also find them useful when attempting to delineatespecies using mitochondrial markers that co-amplifywith nuclear pseudogenes (Numts [27]) Such pseudo-genes are common in many taxa [28-32] and can bevery difficult to tell apart from bona fide mitochondrialsequences [33] However pseudogene sequences divergefollowing speciation just like other markers and repro-ductively isolated species are thus expected to ownmutually exclusive sets of paralogous sequences thatmay not be reciprocally monophyletic but will be easilydetected using haplowebsA previous survey of Pocilloporarsquos molecular diversity

based on a single nuclear marker (ITS2) also includedfour samples from Clipperton [34] in this survey how-ever the authors took the morphological delimitation ofspecies in Veronrsquos Corals of the World [35] as grantedand thus attributed all their samples from Clipperton tothe species ldquoPocillopora effususldquo Even though they pre-sented compelling evidence for the presence of threedivergent ITS2 types among their samples they did notenvision in their article the possibility that cryptic Pocil-lopora species may be present but rather decided toattribute the observed molecular diversity to interspeci-fic hybridization This highlights that molecular studiesbased on a single marker can easily yield erroneous con-clusions especially when a priori morphological hypoth-eses hinder the objective analysis of the molecular dataat hand Moreover this suggests that the current evi-dence for interspecific hybridization in corals will haveto be carefully reevaluated once their species-level tax-onomy becomes clarified

Bipartition scoring vs Doylersquos ml-FFR approachIf analyses using different markers yield different sl-FFRsthen the simplest explanation is that not enough indivi-duals were sequenced resulting in some sl-FFRs that areartefactually smaller than in reality A straightforward wayof solving this discrepancy would be to increase the

number of individuals in the dataset but this cannotalways be done for instance the sampled populations maybe rare or endangered the sampling site may be difficultto access or time and money may be limiting factors Agood example of the consequences of severe undersam-pling can be found in our previously published study ofthe genus Pocillopora in Hawaii [26] in which a firstattempt to delineate graphically sl-FFRs was presented foreach of the four nuclear markers analyzed numeroussmall sl-FFRs were obtained as the number of individualssampled (37 in total) turned out to be very insufficient Inless severe cases however Doylersquos ml-FFR method andorthe bipartition scoring approach presented here can beused to synthesize the results obtained from all markersand obtain a putative species delimitation based on allinformation available at handAlthough Doylersquos ml-FFR procedure may superficially

be mistaken for an approach based on congruence (as itstarts first by analyzing each marker separately to findout sl-FFRs then combines all the results) it is betterdescribed as a ldquototal evidencerdquo [3637] approach sincethe inclusion of a single aberrant dataset can cause thewhole analysis to yield conclusions that are not sup-ported by any other marker A possible way to detectand eliminate such aberrant marker could be to removeone locus at a time and delineate ml-FFRs using theremaining ones but such approach is very time-consuming and will perform poorly if there are severalaberrant datasets Our bipartition-scoring approachhowever solves this problem by making it possible tocompare and quantify the support brought by each mar-ker to the putative species boundariesMissing data when extensive can jeopardize phyloge-

netic analyses [3839] but Doylersquos ml-FFR approach isrelatively immune to this problem since possession of asingle sequence from a known allele pool is a sufficientcriterion for attributing an individual to a given ml-FFR(even when the sequences of this individual for all othermarkers are unavailable) The bipartition scoringapproach presented here however only works if sl-FFRsfor all markers are delineated from the same set ofindividualsWhereas the ml-FFR method proposed by Doyle only

bases itself on the sl-FFRs obtained from nuclear mar-kers our bipartition scoring approach can include othertypes of groupings based on a variety of characters hap-loid sequence markers (such as the mitochondrialregions used in our Pocillopora example) morphologicalcharacters biochemical or immunological propertiesbehavior etc Including a few morphological biochem-ical or behavioral characters in the bipartition scoringanalysis would provide a nice way to test the congru-ence of the patterns obtained from there characterswith those obtained from molecular sequence markers


0 of 14

However our method of quantifying support supposesthat the grouping used (ie all columns in Table 1) beindependent from each other a requirement easily testa-ble in the case of molecular markers (since non-inde-pendent markers are expected to yield congruent genetrees) but that may prove more difficult to establish forother kinds of characters hence one may prefer to usesolely molecular characters for quantifying the supportof the bipartitions

Possible pitfalls dealing with selection shared ancestralsequences and introgressionhybridizationOne possible issue with using the criterion of mutualallelic exclusivity to delineate species is that popula-tions inhabiting contrasting environment can have dis-tinct alleles at loci that are differentially selected forsuch markers under selection one may then end upwith sl-FFRs that are less encompassing than the realspecies as they rather delineate intra-specific ecologicalniches However this problem will only affect markersthat are selected if several markers are used there isgood chance that most of them will be neutral ornear-neutral (or that they will be subjected to differentselective pressures) in which case the congruence ana-lysis presented here will still recover the true speciesboundaries Moreover whenever two sl-FFRs yieldedby a marker turn out to be sympatric populationsinhabiting the same environment then the chance thatthe putative species boundary between them be actu-ally an artefact caused by selection becomes vanish-ingly smallRecent introgression and shared ancestral sequences are

two other possible pitfalls of Doylersquos ml-FFR method theinclusion of a single recently introgressed sequence orshared ancestral haplotype in a dataset supporting other-wise the delimitation of species A and B would cause theimmediate collapse of the two corresponding ml-FFRsinto a single unit and yield the erroneous conclusion thatA and B are conspecific However the bipartition scoringapproach presented here would not be affected by theinclusion of a few such ldquomisbehavingrdquo loci as long as amajority of the markers do behave properly F1 hybridspresent a more difficult problem since they cannot bedetected by a congruence analysis such as our bipartitionscoring approach However if F1 hybrids are relativelyrare in the population (as is usually the case for interspeci-fic hybrids) then they may be spotted in haplowebs asthin lines connecting large haplotype circles (ie infre-quent co-occurrence of common haplotypes which runscontrary to random-mating expectations) If the samesmall set of individuals turns out to be responsible forsuch infrequent connections over nearly all molecularmarkers analyzed one may assume with a high level ofcertainty that these are F1 hybrids and subsequently

remove them from the analysis to ensure proper speciesdelimitation

ConclusionsHaplowebs are versatile tools that combine propertiesfrom both tree-based and non-tree-based approaches tospecies delineation like the former haplowebs can providemeaningful conclusions from relatively few markers with-out relying on population genetic models and like the lat-ter they allow detection of potential species boundaries atan early stage of divergence when populations have notyet reached reciprocal allelic monophyly The methodused for building haplowebs from sets of sequences andfor analyzing the congruence between them is straightfor-ward and reproducible hence our next step will be todevelop programs that speed up the construction of haplo-webs from FASTA sequence alignments and help performthe congruence analysis presented in this article

MethodsSample collection and processingFragments from 74 Pocillopora coral colonies were col-lected while scuba diving or snorkeling on the reefs sur-rounding Clipperton Island (10deg18rsquo00rsquorsquoN 109deg13rsquo00W)from 5 to 14 March 2005 during the international expedi-tion organized by Jean-Louis Etienne Each colonysampled was photographed underwater and its depthrecorded (Table 2) Coral tissues were preserved in buf-fered guanidium thiocyanate solution [4041] and theirDNA purified on an ABI Prism 6100 Nucleic AcidPrepStation

PCR amplification and sequencingWe selectively amplified and sequenced two regions ofthe mitochondrial genome previously identified as themost variable in Pocillopora [22] and two nuclear mar-kers that had been shown to yield useful information onspecies delimitations in this genus [4226] Brieflyamplifications were performed in 25 μl reaction mixescontaining 1x Red Taq buffer 264 μM dNTP 5DMSO 03 μM PCR primers (Table 3) 03 units RedTaq (Sigma) and 10-50 ng DNA PCR conditions com-prised an initial denaturation step of 60 s at 94degC fol-lowed by 40-50 cycles (30 s denaturation at 94degC 30 sannealing at 53degC 75 s elongation at 72degC) and a final5-min elongation step at 72degC PCR products weresequenced in both directions (see primers in Table 3)and sequences were assembled and cleaned usingSequencher 4 (Gene Codes)

Determination of nuclear haplotypesThe ITS2 chromatograms obtained from 31 individualsand the ATPSb chromatograms obtained from 46 indivi-duals contained double peaks indicating that each of these


Page 11 of 14

individuals possessed two sequence types Finding out thesequence types was trivial for 11 pairs of ITS2 chromato-grams that contained only one double peak The ITS2 chro-matogram of 15 individuals and the ATPSb chromatogramof 32 individuals had numerous double peaks as expectedin the case of length-variant heterozygotes [4316] and theirsuperposed sequences were directly deconvoluted using theprogram CHAMPURU [44] (available online at httpwwwmnhnfrjfflotchampuru) The remaining chromatograms

(of 5 individuals for ITS2 and of 14 individuals for ATPSb)had a few double peaks and belonged to heterozygoteswhose allelic sequences were of identical lengths thosewere resolved statistically by reference to the rest of thedataset using SeqPHASE [45] (available online at httpwwwmnhnfrjfflotseqphase) and PHASE [46] All in allthere were 53 distinct multilocus genotypes among our 74samples 46 multilocus genotypes turned out to belong toPocillopora sp A and the 7 others to Pocillopora sp B

Table 2 Localization and depth of each Pocillopora sample collected in Clipperton

Sample name Coordinates Depth (m) Sample name Coordinates Depth (m)

05Clip001 (10deg17rsquo32N 109deg13rsquo34W) 300 05Clip048 (10deg17rsquo07N 109deg12rsquo35W) 110






































Page 12 of 14

Construction of haplonets and haplowebsSequences were aligned in MEGA4 [47] and deposited inpublic databases [GenBankFR729101-FR729473] DNAregions so variable that homology was uncertain wereremoved from the ITS2 and ATPSb alignments Align-ments were converted to the Roehl format using DnaSP[48] median-joining haplotype networks [49] were con-structed using Network 41 (available online at httpwwwfluxus-engineeringcom) and converted intoenhanced metafiles (emf) using Network Publisher(Fluxus Technology) Finally enhanced metafiles wereimported in Microsoft PowerPoint to add colors and con-nections between co-occurring haplotypes (drawn withtheir thickness proportional to the number of individualsin which the said haplotypes were found co-occurring)

Bipartition scoring and reconciliationEach possible bipartition of our 74 samples into two com-plementary sets of individuals was scored according to thenumber of independent datasets supporting it The sup-port of each bipartition was calculated as the percentageof independent datasets that supported this bipartitionA cutoff value of 50 was arbitrarily set to discriminatewell-supported bipartitions from those with little or nosupport and the bipartitions were reconciled in a two-dimensional graph to determine the number of well-supported groups of individuals (ie putative species)among our samples

AcknowledgementsJFF thanks Jean-Louis Etienne for inviting him to take part in his expeditionto Clipperton and Septiegraveme Continent for supporting financially hisparticipation Thanks also to the crew of the Rara Avis for logistic supportduring the expedition to Annie Tillier Josie Lambourdiegravere and Ceacuteline Bonillo(Service de Systeacutematique Moleacuteculaire CNRS UMS 2700 MNHN) for assistancewith lab work and to Guillaume Lecointre and Blaise Li for useful

discussions The remarks of four anonymous reviewers significantly improvedthe manuscript This project was part of agreement ndeg200567 betweenGenoscope and MNHN on the project lsquoMacrophylogeny of lifersquo directed byGuillaume Lecointre support from the Consortium National de Recherche enGeacutenomique is gratefully acknowledged This is contribution ndeg66 from theCourant Research Center ldquoGeobiologyrdquo funded by the German Initiative ofExcellence

Author details1Courant Research Center ldquoGeobiologyrdquo University of GoumlttingenGoldschmidtstr 3 37077 Goumlttingen Germany 2GENOSCOPE Centre Nationalde Seacutequenccedilage 2 rue Gaston Creacutemieux CP5706 91057 Evry Cedex France3UMR UPMC-CNRS-MNHN-IRD 7138 Deacutepartement Systeacutematique et EacutevolutionMuseacuteum National drsquoHistoire Naturelle Case Postale 26 57 rue Cuvier 75231Paris Cedex 05 France 4URBO Department of Biology University of NamurRue de Bruxelles 61 5000 Namur Belgium

Authorsrsquo contributionsJFF devised the method carried out fieldwork DNA extractions and PCRamplifications analyzed the results and drafted the manuscript ACsequenced all PCR products ST supervised the study and revised themanuscript All authors read and approved the final manuscript

Received 15 July 2010 Accepted 30 November 2010Published 30 November 2010

References1 Sites JW Marshall JC Delimiting species a Renaissance issue in

systematic biology Trends in Ecology amp Evolution 2003 18462-4702 Rissler LJ Apodaca JJ Adding more ecology into species delimitation

ecological niche models and phylogeography help define crypticspecies in the black salamander (Aneides flavipunctatus) SystematicBiology 2007 56924-942

3 Wiens JJ Species delimitation new approaches for discovering diversitySystematic Biology 2007 56875-878

4 Savage JM Systematics and the biodiversity crisis BioScience 199545673-679

5 Pfenninger M Schwenk K Cryptic animal species are homogeneouslydistributed among taxa and biogeographical regions BMC EvolutionaryBiology 2007 7121

6 Bickford D Lohman DJ Sodhi NS Ng PKL Meier R Winker K Ingram KKDas I Cryptic species as a window on diversity and conservation Trendsin Ecology amp Evolution 2007 22148-155

7 de Queiroz K The general lineage concept of species species criteriaand the process of speciation A conceptual unification andterminological recommendations In Endless forms Species and speciation

Table 3 Primers used for PCR amplification and sequencing

Marker Primer name Purpose Sequence Reference

ITS2 (nuclear) ITSc2-5 PCR + sequencing 5rsquo-AGCCAGCTGCGATAAGTAGTG-3rsquo [42]

ITS2 (nuclear) R28S1 PCR + sequencing 5rsquo-GCTGCAATCCCAAACAACCC-3rsquo [42]

ATPSβ (nuclear) ATPSβf5 PCR 5rsquo-CCAAGGGTGGNAARATHGGT-3rsquo this article

ATPSβ (nuclear) ATPSβr2 PCR + sequencing 5rsquo-GGTTCGTTCATCTGACCATACAC-3rsquo [26]

ATPSβ (nuclear) ATPSβf2 sequencing 5rsquo-TGAAAGACAAGAGCTCCAAGGTA-3rsquo [26]

ATPSβ (nuclear) ATPSβf4 sequencing 5rsquo-GAGCTGGTGTTGGAAAGACTGT-3rsquo this article

ORF (mitochondrial) FATP61 PCR + sequencing 5rsquo-TTTGGGSATTCGTTTAGCAG-3rsquo [26]

ORF (mitochondrial) RORF PCR + sequencing 5rsquo-SCCAATATGTTAAACASCATGTCA-3rsquo [26]

ORF (mitochondrial) FORF sequencing 5rsquo-GTGCGCCAGCATTCTATTG-3rsquo this article

ORF (mitochondrial) RORF2 sequencing 5rsquo-TAGAATGCTGGCGCACATAA-3rsquo this article

CR (mitochondrial) FNAD52deg PCR + sequencing 5rsquo-GCCYAGRGGTGTTGTTCAAT-3rsquo [26]

CR (mitochondrial) RCOI3deg PCR + sequencing 5rsquo-CGCAGAAAGCTCBARTCGTA-3rsquo [54]

CR (mitochondrial) FNC1 sequencing 5rsquo-GGGGTGAGATGAAGAGGTGA-3rsquo this article

CR (mitochondrial) RNC1 sequencing 5rsquo-CGGGTGCCACTATGTTTTCT-3rsquo this article


Page 13 of 14

Edited by Howard DJ Berlocher SH New York Oxford University Press199857-75

8 Janzen DH Now is the time Philosophical Transaction of the Royal Societyof London Series B 2004 359731-732

9 de Queiroz K Species concepts and species delimitation SystematicBiology 2007 56879-886

10 Funk DJ Omland KE Species-level paraphyly and polyphyly frequencycauses and consequences with insights from animal mitochondrialDNA Annual Review of Ecology Evolution and Systematics 2003 34397-423

11 Doyle JJ The irrelevance of allele tree topologies for speciesdelimitation and a non-topological alternative Systematic Botany 199520574-588

12 Carson H The species as a field for recombination In The species problemEdited by Mayr E Washington American Association for the Advancementof Science 195723-38

13 Miller JT Spooner DM Collapse of species boundaries in the wild potatoSolanum brevicaule complex (Solanaceae S sect Petota) molecular dataPlant Systematics and Evolution 1999 214103-130

14 Marshall JC Areacutevalo E Benavides E Sites JL Sites JW Jr Delimiting speciescomparing methods for Mendelian characters using lizards of theSceloporus grammicus (Squamata Phrynosomatidae) complex Evolution2006 601050-1065

15 Hausdorf B Hennig C Species delimitation using dominant andcodominant multilocus markers Systematic Biology 2010 59491-503

16 Flot J-F Tillier A Samadi S Tillier S Phase determination from directsequencing of length-variable DNA regions Molecular Ecology Notes 20066627-630

17 Harrigan RJ Mazza ME Sorenson MD Computation vs cloning evaluationof two methods for haplotype determination Molecular Ecology Resources2008 81239-1248

18 Palumbi S Baker C Contrasting population structure from nuclear intronsequences and mtDNA of humpback whales Molecular Biology andEvolution 1994 11426-435

19 Miyamoto MM Fitch WM Testing species phylogenies and phylogeneticmethods with congruence Systematic Biology 1995 4464-76

20 Li B Lecointre G Formalizing reliability in the taxonomic congruenceapproach Zoologica Scripta 2009 38101-112

21 Forsman ZH Hunter CL Fox GE Wellington GM Is the ITS region thesolution to the lsquospecies problemrsquo in corals Intragenomic variation andalignment permutation in Porites Siderastrea and outgroup taxaProceedings of the 10th International Coral Reef Symposium 2006 114-23

22 Flot J-F Tillier S The mitochondrial genome of Pocillopora (CnidariaScleractinia) contains two variable regions The putative D-loop and anovel ORF of unknown function Gene 2007 40180-87

23 Posada D Crandall KA Intraspecific gene genealogies trees grafting intonetworks Trends in Ecology and Evolution 2001 1637-45

24 Knowles LL Carstens BC Delimiting species without monophyletic genetrees Systematic Biology 2007 56887-895

25 OrsquoMeara BC New heuristic methods for joint species delimitation andspecies tree inference Systematic Biology 2010 5959-73

26 Flot J-F Magalon H Cruaud C Couloux A Tillier S Patterns of geneticstructure among Hawaiian corals of the genus Pocillopora yield clustersof individuals that are compatible with morphology Comptes RendusBiologies 2008 331239-247

27 Lopez JV Yuhki N Masuda R Modi W OrsquoBrien SJ Numt a recent transferand tandem amplification of mitochondrial DNA to the nucleargenome of the domestic cat Journal of Molecular Evolution 199439174-190

28 Sorenson MD Fleischer RC Multiple independent transpositions ofmitochondrial DNA control region sequences to the nucleus Proceedingsof the National Academy of Sciences of the United States of America 19969315239-15243

29 Bensasson D Zhang D-X Hartl DL Hewitt GM Mitochondrialpseudogenes evolutionrsquos misplaced witnesses Trends in Ecology ampEvolution 2001 16314-321

30 Williams ST Knowlton N Mitochondrial pseudogenes are pervasive andoften insidious in the snapping shrimp genus Alpheus Molecular Biologyand Evolution 2001 181484-1493

31 Richly E Leister D NUMTs in sequenced eukaryotic genomes MolecularBiology and Evolution 2004 211081-1084

32 Schmitz J Piskurek O Zischler H Forty million years of independentevolution a mitochondrial gene and its corresponding nuclearpseudogene in primates Journal of Molecular Evolution 2005 611-11

33 Ibarguchi G Friesen VL Lougheed SC Defeating numts Semi-puremitochondrial DNA from eggs and simple purification methods for field-collected wildlife tissues Genome 2006 491438-1450

34 Combosch DJ Guzman HM Schuhmacher H Vollmer SV Interspecifichybridization and restricted trans-Pacific gene flow in the TropicalEastern Pacific Pocillopora Molecular Ecology 2008 171304-1312

35 Veron JEN Stafford-Smith M Corals of the world Australian Institute ofMarine Science 2000

36 Kluge AG A concern for evidence and a phylogenetic hypothesis ofrelationships among Epicrates (Boidae Serpentes) Systematic Biology1989 397-25

37 Kluge AG Total evidence or taxonomic congruence cladistics orconsensus classification Cladistics 1998 14151-158

38 Sanderson MJ Purvis A Henze C Phylogenetic supertrees assembling thetrees of life Trends in Ecology amp Evolution 1998 13105-109

39 Wiens JJ Missing data incomplete taxa and phylogenetic accuracySystematic Biology 2003 52528-538

40 Sargent TD Jamrich M Dawid IB Cell interactions and the control ofgene activity during early development of Xenopus laevis DevelopmentalBiology 1986 114238-246

41 Fukami H Budd AF Levitan DR Jara J Kersanach R Knowlton NGeographic differences in species boundaries among members of theMontastraea annularis complex based on molecular and morphologicalmarkers Evolution 2004 38324-337

42 Flot J-F Tillier S Molecular phylogeny and systematics of thescleractinian coral genus Pocillopora in Hawaii Proceedings of the 10thInternational Coral Reef Symposium 2006 124-29

43 Creer S Malhotra A Thorpe RS Pook CE Targeting optimal introns forphylogenetic analyses in non-model taxa experimental results in Asianpitvipers Cladistics 2005 21390-395

44 Flot J-F Champuru 10 a computer software for unraveling mixtures oftwo DNA sequences of unequal lengths Molecular Ecology Notes 20077974-977

45 Flot J-F SeqPHASE a web tool for interconverting PHASE inputoutputfiles and FASTA sequence alignments Molecular Ecology Resources 201010162-166

46 Stephens M Smith NJ Donnelly P A new statistical method for haplotypereconstruction from population data The American Journal of HumanGenetics 2001 68978-989

47 Tamura K Dudley J Nei M Kumar S MEGA4 Molecular EvolutionaryGenetics Analysis (MEGA) software version 40 Molecular Biology andEvolution 2007 241596-1599

48 Librado P Rozas J DnaSP v5 a software for comprehensive analysis ofDNA polymorphism data Bioinformatics 2009 251451-1452

49 Bandelt HJ Forster P Roumlhl A Median-joining networks for inferringintraspecific phylogenies Molecular Biology and Evolution 1999 1637-48

50 Hertlein LG Emerson WK Additional notes on the invertebrate fauna ofClipperton Island American Museum novitates 1957 18591-9

51 Glynn PW Veron JEN Wellington GM Clipperton Atoll (eastern Pacific)oceanography geomorphology reef-building coral ecology andbiogeography Coral Reefs 1996 1571-99

52 Carricart-Ganivet JP Reyes-Bonilla H New and previous records of scleractiniancorals from Clipperton Atoll eastern Pacific Pacific Science 1999 53370-375

53 Flot J-F Adjeroud M Les coraux In Clipperton environnement et biodiversiteacutedrsquoun microcosme oceacuteanique Edited by Charpy L Paris Marseille Museacuteumnational drsquoHistoire naturelle IRD 2009155-162

54 Flot J-F Licuanan W Nakano Y Payri C Cruaud C Tillier S Mitochondrialsequences of Seriatopora corals show little agreement with morphologyand reveal the duplication of a tRNA gene near the control region CoralReefs 2008 27789-794

doi1011861471-2148-10-372Cite this article as Flot et al Haplowebs as a graphical tool fordelimiting species a revival of Doylersquos ldquofield for recombinationrdquoapproach and its application to the coral genus Pocillopora inClipperton BMC Evolutionary Biology 2010 10372


Page 14 of 14

Abstract

Background

Results

Conclusions

Background

Results

Nuclear and mitochondrial markers yield putative species-level groupings of individuals

The congruence between independent groupings can be quantified using bipartition scores

Discussion

Using haplowebs to delineate sl-FFRs delimiting species without a priori hypotheses

Bipartition scoring vs Doylersquos ml-FFR approach

Possible pitfalls dealing with selection shared ancestral sequences and introgressionhybridization

Conclusions

Methods

Sample collection and processing

PCR amplification and sequencing

Determination of nuclear haplotypes

Construction of haplonets and haplowebs

Bipartition scoring and reconciliation

Acknowledgements

Author details

Authors contributions

References

species Among the different categories of characters(eg morphological immunological ecological or mole-cular) suitable for assessing the divergence betweenlineages DNA sequences have gained increasing popu-larity among taxonomists in recent years as they are inmany cases not influenced by environmental conditionsnor by the life stage of the organisms under scrutinymoreover gathering information on DNA markersrequires less taxon-specific training and expertise thanfor other categories of characters [8] Methods for deli-miting species based on molecular sequence data can bebroadly classified in two categories tree-based andnon-tree-based [1] Tree-based methods use modelsof sequence evolution to reconstruct nodes that repre-sent hierarchical kinship relationships among organismsin a historical perspective whereas non-tree-basedapproaches use population genetic models to look forevidence of barriers or restrictions to gene flow amongand within extant populationsEven though ldquoa highly corroborated hypothesis of

lineage separation (ie of the existence of separatespecies) requires multiple lines of evidencerdquo [9] as afirst approach it may be desirable to choose a speciesdelimitation criterion that is as general as possible andcan be used systematically Since a recent survey found23 of species reported in the literature to be eitherpoly- or paraphyletic at the various loci investigated[10] a sensitive delimitation method should be capableof detecting closely related species at an early stage oflineage divergence when most of their genetic locihave not yet reached reciprocal monophyly (Figure 1)Moreover it would be advantageous to use as a gen-eral yardstick for delimiting species a method thatdoes not make restrictive assumptions regarding thegenetics of the marker used (eg the absence of copy-number variations) and of the populations studied(eg random mating)One such method applicable to sexually reproducing

organisms was proposed 15 years ago by Doyle [11]briefly this approach uses information on the co-occurrence of alleles in the diploid phase to delineatefor each marker investigated groups of individuals shar-ing a common ldquoallele poolrdquo In Doylersquos terminologythese groups of individuals are named ldquofields for recom-binationrdquo (FFRs) following an earlier proposal by Carson[12] to consider species as groups of individuals whosealleles recombine through segregation and meiosisDoylersquos method relies on the expectation that it takesless time for diverging populations to reach mutualallelic exclusivity than reciprocal allelic monophyly(Figure 1) hence groups of individuals that do not haveany allele in common can be assumed to belong todistinct species even though they are not reciprocallymonophyletic

In Doylersquos original proposal a first step is to delineatefor each genetic locus under scrutiny single-locus FFRs(sl-FFRs) However ldquowith sufficiently fine resolution theallele pool may not extend beyond the single heterozy-gous individuals in which two alleles coexist in suchsituation ldquomany more allele pools are recognized andconcomitantly there are more FFRs each consisting of asingle individualrdquo [11] To overcome this problem Doyleproposes to use information on the co-occurrence ofalleles from different loci to lump sl-FFRs into multilo-cus FFRs (ml-FFRs) if individuals A and B belong to thesame sl-FFR for marker 1 and individuals B and Cbelong to the same sl-FFR for marker 2 then individualsA B and C belong to the same ml-FFR One drawbackof this approach however is that it combines informa-tion from all available markers which makes it difficultto judge the reliability of the ml-FFRs obtained (forinstance the inclusion of a single ancestrally shared orrecently introgressed sequence in a dataset would causethe whole analysis to yield incorrect conclusions) More-over Doylersquos method requires determining the alleles ofmany individuals for several codominant nuclear mar-kers until recently this could only be done for low-resolution andor homoplasy-plagued markers such asallozymes or microsatellites which probably explainswhy earlier attempts to use this method for species deli-mitation were unsuccessful [13-15]In the last few years new techniques have emerged

that make it possible to obtain information-rich allelicsequences from heterozygous individuals without clon-ing [1617] Sequences obtained from exon-primedintrons-crossing (EPIC) nuclear markers [18] as a resultappear now well suited for delimiting species usingDoylersquos method And since such sequences are of finite(and usually moderate) length a simple strategy toobtain sl-FFRs that accurately delineate reproductivelyisolated populations is to sequence a large number ofindividuals One may then assess the reliability of theresulting species boundaries by checking whether theyare supported by several independent molecular mar-kers a congruence approach [1920] that presents theadvantage of not mixing together information from dif-ferent sourcesHere we propose to revive and invigorate Doylersquos

approach by extending it in two directions First wepresent a reliable graphical method to delineate sl-FFRsin large datasets using haplowebs (a contraction of ldquohap-lotype websrdquo) these two-dimensional representations areobtained from conventional haplotype networks (hapl-onetsrdquo) by adding connections between haplotypesfound co-occurring in heterozygous individuals ie hap-lotypes that belong to the same allele pool sensu DoyleAnd second we introduce a procedure to complementthe ml-FFR approach (used by Doyle to infer species


of 14

10 20 30 40 50 60

11 21 31 41 51 61

12 22 32 42 52 62






monophyletic)


speciation

T0 one species



Page 3 of 14












Page 4 of 14



Page 5 of 14

361

603

165

166

97

352

353

224

150

170

605

359

135 316

470

390

470

390

119

265

175

390

470

359

359



05Clip103

05Clip016

05Clip056

361

603

165

166

97

352

353

224

150

170

605

359

135 316

470

390

470

390

119

265

175

390

470

359

359

(a)



Page 6 of 14

394 395 462

147 148

207 216

260 323

40 131

204 212

228 291

357 380

385 396

417 460

473

154

79 373 462

477

40

177

456

226

373 449

142 150

279 401

487

376 449 285

449 376 40

396

1 106

122 123

200

154

178 224 319 385 455

125 370 435

145 374

40 376

129 208

207 301

129 224 279 280 342

234 280

385

455 473

475

483

120 120 120

172 373

394 395 462

147 148

207 216

260 323

40 131

204 212

228 291

357 380

385 396

417 460

473

154

79 373 462

477

40

177

456

226

373 449

142 150

279 401

487

376 449 285

449 376 40

396

1 106

122 123

200

154

178 224 319 385 455

125 370 435

145 374

40 376

129 208

207 301

129 224 279 280 342

234 280

385

455 473

475

483

120 120 120

172 373


05Clip102


05Clip103

(a)

(b)



Page 7 of 14




119

148

191

205

218

235

266

271

314

322

345

356

363

629

655

1810

1455

1447



05Clip048 05Clip056




















Page 8 of 14



05Clip096 05Clip103




05Clip048


05Clip016

05Clip056

1

2

3

4

5

6

7

8

9

10

11

1213



05Clip103




1

(a)

(b)

Pocillopora sp A



Page 9 of 14










Page 10 of 14










Page 11 of 14











































Page 12 of 14

































Page 13 of 14



















































Page 14 of 14

Abstract

Background

Results

Conclusions

Background

Results



Discussion




Conclusions

Methods






Acknowledgements

Author details


References

10 20 30 40 50 60

11 21 31 41 51 61

12 22 32 42 52 62






monophyletic)


speciation

T0 one species



of 14












Page 4 of 14



Page 5 of 14

361

603

165

166

97

352

353

224

150

170

605

359

135 316

470

390

470

390

119

265

175

390

470

359

359



05Clip103

05Clip016

05Clip056

361

603

165

166

97

352

353

224

150

170

605

359

135 316

470

390

470

390

119

265

175

390

470

359

359

(a)



Page 6 of 14

394 395 462

147 148

207 216

260 323

40 131

204 212

228 291

357 380

385 396

417 460

473

154

79 373 462

477

40

177

456

226

373 449

142 150

279 401

487

376 449 285

449 376 40

396

1 106

122 123

200

154

178 224 319 385 455

125 370 435

145 374

40 376

129 208

207 301

129 224 279 280 342

234 280

385

455 473

475

483

120 120 120

172 373

394 395 462

147 148

207 216

260 323

40 131

204 212

228 291

357 380

385 396

417 460

473

154

79 373 462

477

40

177

456

226

373 449

142 150

279 401

487

376 449 285

449 376 40

396

1 106

122 123

200

154

178 224 319 385 455

125 370 435

145 374

40 376

129 208

207 301

129 224 279 280 342

234 280

385

455 473

475

483

120 120 120

172 373


05Clip102


05Clip103

(a)

(b)



Page 7 of 14




119

148

191

205

218

235

266

271

314

322

345

356

363

629

655

1810

1455

1447



05Clip048 05Clip056




















Page 8 of 14



05Clip096 05Clip103




05Clip048


05Clip016

05Clip056

1

2

3

4

5

6

7

8

9

10

11

1213



05Clip103




1

(a)

(b)

Pocillopora sp A



Page 9 of 14










Page 10 of 14










Page 11 of 14











































Page 12 of 14

































Page 13 of 14



















































Page 14 of 14

Abstract

Background

Results

Conclusions

Background

Results



Discussion




Conclusions

Methods






Acknowledgements

Author details


References












of 14



Page 5 of 14

361

603

165

166

97

352

353

224

150

170

605

359

135 316

470

390

470

390

119

265

175

390

470

359

359



05Clip103

05Clip016

05Clip056

361

603

165

166

97

352

353

224

150

170

605

359

135 316

470

390

470

390

119

265

175

390

470

359

359

(a)



Page 6 of 14

394 395 462

147 148

207 216

260 323

40 131

204 212

228 291

357 380

385 396

417 460

473

154

79 373 462

477

40

177

456

226

373 449

142 150

279 401

487

376 449 285

449 376 40

396

1 106

122 123

200

154

178 224 319 385 455

125 370 435

145 374

40 376

129 208

207 301

129 224 279 280 342

234 280

385

455 473

475

483

120 120 120

172 373

394 395 462

147 148

207 216

260 323

40 131

204 212

228 291

357 380

385 396

417 460

473

154

79 373 462

477

40

177

456

226

373 449

142 150

279 401

487

376 449 285

449 376 40

396

1 106

122 123

200

154

178 224 319 385 455

125 370 435

145 374

40 376

129 208

207 301

129 224 279 280 342

234 280

385

455 473

475

483

120 120 120

172 373


05Clip102


05Clip103

(a)

(b)



Page 7 of 14




119

148

191

205

218

235

266

271

314

322

345

356

363

629

655

1810

1455

1447



05Clip048 05Clip056




















Page 8 of 14



05Clip096 05Clip103




05Clip048


05Clip016

05Clip056

1

2

3

4

5

6

7

8

9

10

11

1213



05Clip103




1

(a)

(b)

Pocillopora sp A



Page 9 of 14










Page 10 of 14










Page 11 of 14











































Page 12 of 14

































Page 13 of 14



















































Page 14 of 14

Abstract

Background

Results

Conclusions

Background

Results



Discussion




Conclusions

Methods






Acknowledgements

Author details


References



of 14

361

603

165

166

97

352

353

224

150

170

605

359

135 316

470

390

470

390

119

265

175

390

470

359

359



05Clip103

05Clip016

05Clip056

361

603

165

166

97

352

353

224

150

170

605

359

135 316

470

390

470

390

119

265

175

390

470

359

359

(a)



Page 6 of 14

394 395 462

147 148

207 216

260 323

40 131

204 212

228 291

357 380

385 396

417 460

473

154

79 373 462

477

40

177

456

226

373 449

142 150

279 401

487

376 449 285

449 376 40

396

1 106

122 123

200

154

178 224 319 385 455

125 370 435

145 374

40 376

129 208

207 301

129 224 279 280 342

234 280

385

455 473

475

483

120 120 120

172 373

394 395 462

147 148

207 216

260 323

40 131

204 212

228 291

357 380

385 396

417 460

473

154

79 373 462

477

40

177

456

226

373 449

142 150

279 401

487

376 449 285

449 376 40

396

1 106

122 123

200

154

178 224 319 385 455

125 370 435

145 374

40 376

129 208

207 301

129 224 279 280 342

234 280

385

455 473

475

483

120 120 120

172 373


05Clip102


05Clip103

(a)

(b)



Page 7 of 14




119

148

191

205

218

235

266

271

314

322

345

356

363

629

655

1810

1455

1447



05Clip048 05Clip056




















Page 8 of 14



05Clip096 05Clip103




05Clip048


05Clip016

05Clip056

1

2

3

4

5

6

7

8

9

10

11

1213



05Clip103




1

(a)

(b)

Pocillopora sp A



Page 9 of 14










Page 10 of 14










Page 11 of 14











































Page 12 of 14

































Page 13 of 14



















































Page 14 of 14

Abstract

Background

Results

Conclusions

Background

Results



Discussion




Conclusions

Methods






Acknowledgements

Author details


References

361

603

165

166

97

352

353

224

150

170

605

359

135 316

470

390

470

390

119

265

175

390

470

359

359



05Clip103

05Clip016

05Clip056

361

603

165

166

97

352

353

224

150

170

605

359

135 316

470

390

470

390

119

265

175

390

470

359

359

(a)



of 14

394 395 462

147 148

207 216

260 323

40 131

204 212

228 291

357 380

385 396

417 460

473

154

79 373 462

477

40

177

456

226

373 449

142 150

279 401

487

376 449 285

449 376 40

396

1 106

122 123

200

154

178 224 319 385 455

125 370 435

145 374

40 376

129 208

207 301

129 224 279 280 342

234 280

385

455 473

475

483

120 120 120

172 373

394 395 462

147 148

207 216

260 323

40 131

204 212

228 291

357 380

385 396

417 460

473

154

79 373 462

477

40

177

456

226

373 449

142 150

279 401

487

376 449 285

449 376 40

396

1 106

122 123

200

154

178 224 319 385 455

125 370 435

145 374

40 376

129 208

207 301

129 224 279 280 342

234 280

385

455 473

475

483

120 120 120

172 373


05Clip102


05Clip103

(a)

(b)



Page 7 of 14




119

148

191

205

218

235

266

271

314

322

345

356

363

629

655

1810

1455

1447



05Clip048 05Clip056




















Page 8 of 14



05Clip096 05Clip103




05Clip048


05Clip016

05Clip056

1

2

3

4

5

6

7

8

9

10

11

1213



05Clip103




1

(a)

(b)

Pocillopora sp A



Page 9 of 14










Page 10 of 14










Page 11 of 14











































Page 12 of 14

































Page 13 of 14



















































Page 14 of 14

Abstract

Background

Results

Conclusions

Background

Results



Discussion




Conclusions

Methods






Acknowledgements

Author details


References

394 395 462

147 148

207 216

260 323

40 131

204 212

228 291

357 380

385 396

417 460

473

154

79 373 462

477

40

177

456

226

373 449

142 150

279 401

487

376 449 285

449 376 40

396

1 106

122 123

200

154

178 224 319 385 455

125 370 435

145 374

40 376

129 208

207 301

129 224 279 280 342

234 280

385

455 473

475

483

120 120 120

172 373

394 395 462

147 148

207 216

260 323

40 131

204 212

228 291

357 380

385 396

417 460

473

154

79 373 462

477

40

177

456

226

373 449

142 150

279 401

487

376 449 285

449 376 40

396

1 106

122 123

200

154

178 224 319 385 455

125 370 435

145 374

40 376

129 208

207 301

129 224 279 280 342

234 280

385

455 473

475

483

120 120 120

172 373


05Clip102


05Clip103

(a)

(b)



of 14




119

148

191

205

218

235

266

271

314

322

345

356

363

629

655

1810

1455

1447



05Clip048 05Clip056




















Page 8 of 14



05Clip096 05Clip103




05Clip048


05Clip016

05Clip056

1

2

3

4

5

6

7

8

9

10

11

1213



05Clip103




1

(a)

(b)

Pocillopora sp A



Page 9 of 14










Page 10 of 14










Page 11 of 14











































Page 12 of 14

































Page 13 of 14



















































Page 14 of 14

Abstract

Background

Results

Conclusions

Background

Results



Discussion




Conclusions

Methods






Acknowledgements

Author details


References




119

148

191

205

218

235

266

271

314

322

345

356

363

629

655

1810

1455

1447



05Clip048 05Clip056




















of 14



05Clip096 05Clip103




05Clip048


05Clip016

05Clip056

1

2

3

4

5

6

7

8

9

10

11

1213



05Clip103




1

(a)

(b)

Pocillopora sp A



Page 9 of 14










Page 10 of 14










Page 11 of 14











































Page 12 of 14

































Page 13 of 14



















































Page 14 of 14

Abstract

Background

Results

Conclusions

Background

Results



Discussion




Conclusions

Methods






Acknowledgements

Author details


References



05Clip096 05Clip103




05Clip048


05Clip016

05Clip056

1

2

3

4

5

6

7

8

9

10

11

1213



05Clip103




1

(a)

(b)

Pocillopora sp A



of 14










Page 10 of 14










Page 11 of 14











































Page 12 of 14

































Page 13 of 14



















































Page 14 of 14

Abstract

Background

Results

Conclusions

Background

Results



Discussion




Conclusions

Methods






Acknowledgements

Author details


References










of 14










Page 11 of 14











































Page 12 of 14

































Page 13 of 14



















































Page 14 of 14

Abstract

Background

Results

Conclusions

Background

Results



Discussion




Conclusions

Methods






Acknowledgements

Author details


References










of 14











































Page 12 of 14

































Page 13 of 14



















































Page 14 of 14

Abstract

Background

Results

Conclusions

Background

Results



Discussion




Conclusions

Methods






Acknowledgements

Author details


References











































of 14

































Page 13 of 14



















































Page 14 of 14

Abstract

Background

Results

Conclusions

Background

Results



Discussion




Conclusions

Methods






Acknowledgements

Author details


References

































of 14



















































Page 14 of 14

Abstract

Background

Results

Conclusions

Background

Results



Discussion




Conclusions

Methods






Acknowledgements

Author details


References



















































of 14

Abstract

Background

Results

Conclusions

Background

Results



Discussion




Conclusions

Methods






Acknowledgements

Author details


References

Haplowebs as a graphical tool for delimiting species: a revival of

Documents

Transcript of Haplowebs as a graphical tool for delimiting species: a revival of