Olive domestication from structure of oleasters and cultivars using nuclear RAPDs and mitochondrial RFLPs

HAL Id: hal-02679443

https://hal.inrae.fr/hal-02679443

Submitted on 31 May 2020

HAL is a multi-disciplinary open access

archive for the deposit and dissemination of

sci-entific research documents, whether they are

pub-lished or not. The documents may come from

teaching and research institutions in France or

abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est

destinée au dépôt et à la diffusion de documents

scientifiques de niveau recherche, publiés ou non,

émanant des établissements d’enseignement et de

recherche français ou étrangers, des laboratoires

publics ou privés.

Olive domestication from structure of oleasters and

cultivars using nuclear RAPDs and mitochondrial

RFLPs

G. Besnard, Philippe Baradat, C. Breton, Bouchaïb Khadari, A. Bervillé

To cite this version:

Original article

Olive domestication from structure

of oleasters and cultivars using nuclear

RAPDs and mitochondrial RFLPs

GuillaumeBesnard, PhilippeBaradat, Catherine Breton,

Bouchaib Khadari, André Bervillé

∗

UnitédeRecherchesGénétiqueetAméliorationdesPlantes,Institutnationaldela

rechercheagronomique,Bât. 33,2placeP.Viala,34060MontpellierCedex1,France

Abstract RAPDprolesof121olivecultivarswerecomparedtothoseof20

natu-raloleasterpopulationsfromeasternandwesternpartsoftheMediterraneanBasin.

ConsideringtheproximitiesofRAPDprolesbetweencultivarsandeasternorwestern

oleaster populations, clear dierencesappearedbetween groupsof cultivars.

Culti-varsfromIsrael,Turkey,Syria,GreeceandSicilywere,withveryfewexceptions,close

to theeastern groupof oleasters; incontrast, clones from Continental Italy,

Conti-nental France, Corsica, Spain and the Maghreb were closer to the western group.

This genetic structureis coherent with alocal selection of cultivarsall around the

Mediterranean Basin. Thecultivarswerealso characterised for their mitochondrial

cytotype. Thisinformationledtotheconclusionthata greatmajority(103of121)

ofthecultivarsoriginatedbymaternaldescentfromtheeasternpopulationsasthey

carry the mitotypes ME1or ME2. However, the selection process, which involved

hybridisationby pollen from local populations,could haveled toan RAPDprole

closertowesternthantoeasternnaturalpopulations. Furthermore,theother

culti-varswiththewesternmitotypesMOMorMCKgenerallykeptanuclearRAPDprole

close tothe proleofwesternnatural populations. Consequently,theycouldresult

fromexclusivelylocalmaterial(asfor Corsica). Cultivarsdisplayingsuchmitotypes

could alsohave beenselectedinhybrids orintrogressedgenotypesbetween western

local oleasters and the introduced eastern varietiesused as male parents,restoring

an easternRAPD type. Therefore,the process ofolive domesticationappeared as

disymmetrical: thewesternMediterraneanisprobablyazonewhereolivetreesfrom

theEast,onceintroduced,havebeenhybridisedandback-crossedwiththeindigenous

olives.

domestication/geneticstructure/mitotype/Olea europaea /oleaster/

olive

∗

Résumé Domestication de l'olivier d'aprèsla structure descultivars et

des oléastres basée sur les marqueurs nucléaires RAPD et les RFLP de

l'ADN mitochondrial. Les prolsRAPDde 121cultivars d'olivieront été

com-parés à ceux de 20 populations naturelles d'oléastres de l'Est et de l'Ouest de la

Méditerranée. EnconsidérantlesproximitésdesprolsRAPDentreles cultivarsde

l'Est et de l'Ouest et les populations, des diérences nettes apparaissent entre les

groupesdecultivars. Lescultivarsd'Israël,Turquie,Syrie,GrèceetSicilesont,avec

peu d'exceptions,prochesdes oléastres de l'Est; à l'opposé les clones d'Italie

con-tinentale,de France continentalede Corse, d'Espagneetdu Maghreb sont proches

des oléastres del'Ouest. Cettestructuregénétiqueest cohérenteavecunesélection

localedescultivarstoutautourduBassinMéditerranéen. Lescultivarsontaussiété

caractérisés parlecytotypemitochondrial. Cetteinformationmontrequelagrande

majorité(103sur121)descultivarsportelesmitotypesME1ouME2,etauraitdonc

une origine maternelle dans les populations de l'Est. Cependant, le processus de

sélection quiimpliquel'hybridationparlepollendespopulationslocales,pourraient

avoirunprolRAPDplusprochedespopulationsdel'Ouestquedespopulations

na-turellesdel'Est. Enoutre,lesautrescultivars,aveclesmitotypesdel'OuestMOMou

MCK,ont généralementconservéunprol RAPDprochedespopulationsnaturelles

del'Ouest. Enconséquence,ellespourraientrésulterdeformesexclusivementlocales

(commeenCorse). Cependant,lescultivarsquiportentdetelsmitotypespourraient

aussiavoirétésélectionnésàpartird'hybridesoudegénotypesintrogressésentreles

oléastresdel'Ouestetlesvariétés,introduitesdel'Est,utiliséescommeparentmâle,

conduisant à un prolRAPD de l'Est. Le processus de domestication del'olivier

apparaîtdissymétrique:laMéditerranéedel'Ouestseraitunezoneoùlesoliviersde

l'Est,unefoisintroduits,auraitétéhybridésetrétro-croisésaveclesoliviersindigènes.

domestication/structuregénétique/mitotype/Olea europaea/oléastre/

olive

Abbreviations. ME1 = Mitotype eastern Mediterranean n

◦

1; ME2 =

Mitotype eastern Mediterranean n

◦

2; MOM =Mitotypewestern

Mediter-ranean; MCK = Mitotype characteristic of the cultivarChemlal de Kabylie

(westernMediterranean).

1. INTRODUCTION

In fruit trees, knowledge about the structure of genetic diversity is of a

great interest for the management of genetic resources. Characterisationof

the genetic diversityin bothwild and cultivated forms is necessaryto study

similaritiesbetweenthemandtodeterminethemostlikelyoriginsofcultivars.

Thisapproachshouldleadtoabetterunderstandingofthedomestication

pro-cess: localisationofthedomesticationcentresandevidenceofhybridisationor

introgression. From morphologicaldata, ithasbeensupposed that

hybridisa-tionbetweencultivated formsandtheirwild relativesoccurredin severalfruit

ofhybridisationsbetweendierentiatedtaxa,asinaapple[9,12],incacao[17],

andin coee[15].

Theolivetree is associated withthe historyof theMediterranean

civilisa-tions. It is widely assumed that it has been exploited for at least nineteen

thousandyears[15]. Usesofoleasters(wild olive)during theNeolithic period

in dierent parts of the Mediterranean Basin (Near-East, Spain) have been

reportedbydierentauthors[29,33]. Thedomesticationofthespecies, which

implied mainly the establishment of orchards and the vegetative

multiplica-tion of the agronomically interesting trees, is supposed to have occurred in

the Near-Eastbyabout5,7005,500yearsago[34]. Thecultivars could have

beendisseminatedtowardsthe MediterraneanBasinfrom theNearEast

dur-ing human migrations. Elant [10] has supposed that, in the western partof

theMediterranean,hybridisationbetweenindigenousoleasters andcultivated

formsintroducedfromtheLevantcouldhaveledtonewcultivatedforms.

Oleasters areconsidered asindigenous to theMediterranean Basin[34] up

to SouthernFrance[26]. It isassumed thatcultivars havederivedfrom some

Mediterranean wild populations, but it is also supposed that some oleasters

derivedfrom culture(feral forms) or couldbe hybrids with cultivated forms.

Thus, thestudy ofthegeneticstructure ofoleasters should takeintoaccount

allinformationandhypotheses. Aclosegeneticproximitybetweenoleasterand

cultivated olivehasalreadybeenshown withisozymes [18]and RAPDs[4,6].

Wehaveshownthatsomecultivarscouldhavebeenselectedfromoleasterinthe

westernMediterranean[5]. Thus,theoriginofcultivated olivehasbeen

com-plex andseveral domesticationcentres mayhaveexisted. Furthermore,using

nuclear andmitochondrialDNAmarkers,ahigh geographicgeneticstructure

wasalso shown for oleasters[4]. Two main groupscanbe distinguished: the

Eastern populations characterised by the mitotype ME1, and western

popu-lations characterised by the mitotypes MOM and MCK. This structure has

probably resulted from the regression of the natural populations during the

last glaciations leadingtoa highgenetic drift. Incontrast, theferalforms in

thewesternMediterranean wereoftencharacterisedbythemitotypeME1[4].

Inthe presentstudy, weused boththe nuclearRAPDprolesfor 57

poly-morphic markersand theRFLP mitotypesto obtaincomplementary

informa-tionduetothedierentmodesofinheritance(bi-parentalorpurelymaternal).

Weassessedthegeneticproximitybetweennaturalpopulations andgroupsof

cultivars. Since cultivars wererelated to dierent populations, this sustained

multilocal selectionfor cultivars. Moreover,wealso found cultivars probably

2. MATERIAL ANDMETHODS

2.1. Vegetal material

Cultivars of reference(codes1to 102) and 19cultivated forms with

unde-termined denominations (codes 103 to 121) were characterised with RAPDs

(Tab.I)[5]. Twentyoleasterpopulationswereprospectedaroundthe

Mediter-ranean Basin(Tab. II) [6]. Thesepopulationswere sampledasfar aspossible

frompresentorchards. Weassumedthatthetreeswerenaturallydisseminated

becausetheywerenotalignedandwereusuallyin areasdiculttoaccess. A

total of 235 oleasters was studied representing most of the countries (except

someIslands). Due toregressionofsomepopulations,thenumberoftreesper

population varied from 5 to 22. In this sample, populations from the East

carried the ME1 mitotype whereas western populations most frequently

car-ried MOMand MCKmitotypes[4]. Incontrast, aweakstructure wasshown

forcultivars [5]. This arguesforawildoriginofthesampled oleaster

popula-tions. Nevertheless,wecannotexcludethehypothesisthatgeneowsfromthe

cultivatedforms haveintrogressedoleasters.

2.2. Moleculardata

DNA preparation was previously described by Besnard et al. [3]. The

RAPD amplication and electrophoresis procedures were described by

Quil-letetal. [23]. Eightprimers(Bioprobe,France)wereusedontheDNAsfrom

all individuals: A1, A2, A9, A10, C9, C15, E15, O8. All thetrees were also

characterisedformitochondrialDNApolymorphismusingtheRFLPmethodas

describedbyBesnardetal.[3]. Thetreeswereclassiedaccordingtofour

mi-totypescalledME1,ME2,MOMandMCK[3]. Geographicspecicityofthese

mitotypeshas beenpreviouslydiscussed by Besnard and Bervillé[4].

Never-theless, themitotype ME2 was notfound in oleasters. CpDNA and mtDNA

polymorphismassociation hassuggestedthat thecytotypeME2 marksa

dis-tinct oleasterpopulationthat veryprobably emergedfrom theEast

Mediter-ranean, in aregion which wasnotstudied (i.e. North-eastSyria,Cyprusand

Crete) (Besnard et al. unpublished data). We therefore considered this

mi-totype asoriginating fromthe easternMediterraneansince itwasrevealed in

severalcultivars typicalofthisregion(i.e., Zaity,ToahiandAmygdalolia).

2.3. Statisticalanalyses

The statistical analyses were performed with the OPEP software [1]. We

usedadiscriminantanalysisonqualitativedataofthe20naturalpopulations,

following the model described by Saporta [25] and Lebart et al. [16] as the

DISQUAL procedure. The principle of this method is to compute the axes

Table II. List of oleaster populations analysed, location and size of the samples.

Geographicgroups(East,Westorintermediary)weredenedaccordingtothegenetic

structurerevealedbyBesnardetal.[6].

N

◦

pop Localisationofoleasters Country Size East/West

1 Urla,Izmir Turkey 6 East

2 Harim,Orontevalley Syria 13 East

3 AlAscharinah,ElGhab Syria 12 East

4 MontCarmel,Haifa Israel 18 East

5 Cyrenaique Libya 15 Intermediateprole

6 Zaghouan Tunisia 6 West

7 MontBelloua,TiziOuzou Algeria 7 West

8 Tamanar,Essaouira Morocco 10 West

9 Immouzzer,HighAtlas Morocco 5 West

10 Torviczon,Andalucia Spain 15 West

11 Asturias Spain 5 West

12 CapdesMèdes,Porquerolles,Var France 22 West

13 LaRepentence,Porquerolles,Var France 12 West

14 MontBoron,Nice,AlpesMaritimes France 22 West

15 Ostricone,Corsica France 11 West

16 Ogliastro,Corsica France 10 West

17 Filitosa,Corsica France 11 West

18 Bonifacio,Corsica France 6 West

19 Messine,Sicily Italy 12 West

20 Ali,Sicily Italy 17 West

individuals and then to use the individual coordinates on all the axes or on

the most signicant of them in a discriminant analysis. If

p < n

there are

p − 1

axes; if

p > n

there are

n − 1

axes. Forthis application, with 56 axes

from 57 polymorphic RAPDs, we retained the 33 axes for which the

F

test

ofpopulationeectwassignicantatleastat the5%level. The121cultivars

wereusedassupplementarydataandprojectedontheaxesofthediscriminant

analysis onlybuiltfrom theRAPDdataofnaturalpopulations. Normalityof

distributionsofthecoordinateswerecheckedaswellashomoscedasticityofthe

within-populationvariances(Bartlett'stest). TheKullbacktestofhomogeneity

ofvariance-covariancematricescouldnotbeperformedbecausethesizesofthe

populationsweretoolimitedcomparedwith thenumberofmarkers.

The cultivars were characterised by their relative distances to the general

centroidsofeasternandwesternnaturalpopulations. Asyntheticnew

param-eter was built for the RAPD prole of each cultivar, to measure its relative

average proximity to eastern or western oleasters. This parameter,

D

i

, was

where

D

2

_(G

i

, G

East

)

istheMahalanobisdistance,inthemultidimensionalspace,

betweenthe

i

th

cultivar,

G

i

,andthecentroidofallthenaturalindividualsfrom

theeasterngroup.

G West

H

GEast

u

v

s

t

r

Gi

The other parameter:

D

2

_(G

i

, G

West

)

is the corresponding distance of

G

i

fromthewesterngroup. Wefurthername

r

and

s

thesquarerootsofthe

dis-tancesofthecultivarfromtheeasternandwesternnaturalpopulations,

respec-tively. Theyarethecorrespondingdeviations. Asshownbytheabovedrawing,

thisparameterreplacesanunbiased(butnotestimablehere)expressionofthe

relativedeviation fromEast,

D

0

i

= 100

u+v

v

, where

u

and

v

arethedeviations

from the projection of

G

i

,

H

, on the straight line joining

G

West

and

G

East

.

When thedeviation,

t

, between

G

i

and

H

increases

D

i

= 100

√

r

2

_−t

2

√

s

2

_−t

2

₊

√

_r

2

_−t

2

may become a strongly biased estimate of

D

0

i

because it tends towards the

limit value:

D

i

= 100

√

r

2

_−t

2

2

√

r

2

_−t

2

= 50

when

t

is large, leading to a ratio

r/s

close to 1; but this parameter, although theoretically less sensitive than

D

0

i

,

givesqualitativelythesameinformationas

D

0

i

,withthesamecriticalvalueof

50 for delineation betweentwocategories of olivesaccording to their genetic

proximityto easternand western naturalpopulations. For

D

i

, indeed, asfor

D

0

i

, thevalueslowerthan50indicate that thecultivariscloserto theeastern

populationsand thevaluesgreaterthan50characterisethecultivarscloserto

thewesternpopulations,asavalueof50isexpectedforthetreesexactly

inter-mediatebetweenthesetwogroups. WehaveexcludedthenaturalLibyanolive

treestodenethetwocentroids,becausetheirRAPDprolewasshowntobe

intermediatebetweenEastandWest (seebelow). Forthe33inputvariatesof

thediscriminantanalysisofthe20naturalpopulationsassupplementarydata

togetherwithindividualvaluesofthecultivars,thethreeparameters

s

,

r

,and

D

,wereestimatedforeachofthe121RAPDprolesbyusingthemeansofthe

western and eastern populations. These three parameters were then used as

a visualisation of thenew cultivar clustering) and the usual additionaltests,

derivedfromvarianceanalysis,whichwillbedescribedfurther. Thereafterthe

D

parameterwill becalledfurther East-Westratio.

3. RESULTS

3.1. Generaloverview

Thediscriminantanalysisofthe20naturalpopulationsgave11axes

signi-cantlydiscriminantatthe5%level(Wilkstests). Theseaxescumulated92.6%

ofthediscrimination. Allthebetween-populationMahalanobisdistanceswere

signicantatleastatthe0.1%levelexceptforonecomparison,betweenMont

BoroninFrance(n

◦

18)andMontBellouainAlgeria(n

◦

7)(resultsnotshown).

This dierencewashoweversignicantat the10%level. Themain featureof

the structure of these populationswas apparent on the rst twoaxes

cumu-lating 43.38% ofthe discrimination asdisplayed byFigure 1A. Therst axis

clearlyseparatedtheeasterngroupof naturalpopulations (Turkey,Syriaand

Israel)fromthewesterncluster(Corsica,Sicily,ContinentalFrance,Spain,

Al-geriaand Tunisia). TheuniqueLibyan populationappearedasintermediate.

This fact iscoherentwiththegeographicsituation ofthis country;weshould

notethat,amongthe19otheroleasterpopulations,Tunisia,clusteredintothe

westerngroup,istheclosestto Libya.

Considerationoftheprojectionsofthe121cultivarsidentiedbytheir

mito-typeshowedthatthecultivars displayingthemitotypesMOMandMCKwere

generally far from the four populations representative of the eastern group

(Turkey,SyriaandIsrael)andclosetoawesternoleasterpopulation(Fig.1A).

These mitotypes havebeen shown to be specic to the western oleasters [4].

Themoststrikingfeature wasthatthemitotypesME1 andME2, cumulating

103 trees, appeared as much more homogeneouslydistributed, with smallest

distancesaslikelyfrompopulationsofwestern,easternorintermediategroups.

Figure1Bshowstheclusteringofthe121cultivarsrelativetotheircountry

oforigin. Themainfeatureisthenon-randomdistribution ofcultivatedolives

ofagivencountry. Thecultivarsdisplayamoreorlesscloseproximitywiththe

naturalpopulationsfrom their countryorregion orfrom alarger geographic

area. ThisisespeciallythecaseforCorsicanandAlgerianvarietieswithin the

western countries and for Turkish, Syrian or Israeli olives within the eastern

-0,4

-0,3

-0,2

-0,1

0

0,1

0,2

0,3

-0,6

-0,5

-0,4

-0,3

-0,2

-0,1

0

0,1

0,2

0,3

Axis 1 (31.99% of the discrimination)

Axi

s

2

(11.

39%

of

th

e

di

scr

im

in

at

io

n)

Corsica

Sicily

Continental France

Spain

Morocco

Algeria

Tunisia

Lybia

Turkey

Syria

Israel

Cultivars of ME1 mitotype

Cultivars of ME2 mitotype

Cultivars of MOM mitotype

Cultivars of MCK mitotype

A

-0,4

-0,3

-0,2

-0,1

0

0,1

0,2

0,3

-0,6

-0,5

-0,4

-0,3

-0,2

-0,1

0

0,1

0,2

0,3

Axis 1 (31.99% of the discrimination)

A

xis

2

(11.39%

of

th

e

discrimination)

w.Corsica

c.Corsica

w.Sicily

c.Sicily

w.Continental France

c.Continental France

w.Spain

c.Spain

w.Morocco

c.Morocco

w.Algeria

c.Algeria

w.Tunisia

c.Tunisia

w.Turkey

c.Turkey

w.Syria

c.Syria

w.Israel

c.Israel

w.Lybia

c.Greece

c.Lebanon

c.Egypt

c.Yugoslavia

c.Continental Italy

c.Portugal

B

Figure 1. Genetic proximities between 121cultivars and 20 oleaster populations.

A. The cultivars are identied by their mitotype. B. The cultivars are identied

bytheircountryorregionof origin. Inthelegend, thecultivarsare prexedbyc

andtheoleasterpopulationsbyw. Thecentroidsoftheseoleasterpopulationsare

3.2. Further examination of the relation between relatedness

ofcultivarstonatural populationsand theircountryoforigin

Thesecond discriminantanalysisof countryorregioneect, performed on

s

,

r

and

D

,usedonlythe11countriesorregionsrepresentedbyseveral

culti-vars(from4to44proles)tobuildthenewaxes(thecultivarsfrom Morocco,

Lebanon, Egypt, Yugoslavia and Portugal were only used as supplementary

data). Only one axiswas signicantly discriminantat the0.1% level (Wilks

test)with83.43%ofthediscrimination. Figure2Apresentsthepositionsofthe

countryorregioncentroidsandoftheindividualcultivarsonthersttwoaxes.

Themostimportantresultisthattheonlydiscriminantaxis(axis1)appeared

asquiteperfectlycorrelatedwiththeEast-Westratio(

r = 0.998

∗∗∗

). The

pre-vailingvalueof thisparameterfor cultivarstructurepercountryisconrmed

byTableIIIwhichdisplaystherelationshipbetweencountryorregionranking

for

D

and theircorrespondingdistributiononaxis1ofthediscriminant

anal-ysis withthe associated Newman-Keulstests. This close correspondence was

measuredbyahigh valueof theSpearmanrankcorrelation(

r = 0.982

∗∗∗

). It

meansthattheonlyfactortobeconsideredforcharacterisationofthecultivar

proximitytothenaturalpopulationsofMediterraneanBasinisnottheabsolute

genetic distance to these populations (expressedby the

s

and

r

parameters),

buttherelativeproximitiestothetwogroupsthatwehavecharacterised.This

fact hasveryprofoundconsequences,which willbediscussedfurther.

ThevalidationoftheEast-Westratioasanadequatemeasureofthegenetic

background ofoleasters in the cultivars isimportant to classifythe countries

and the individual cultivars for predominanceof the western or easterngene

pool. Therefore,TableIIImaybemoreeasilyinterpreted: cultivarsfromIsrael,

Greece, Turkey, Syriaand Sicily haveon theaveragea prevailing relatedness

with easternoleasters, whereascultivars from Corsica, Algeriaand Spainare

mostlyrelatedto western oleasters. Cultivars from Tunisia, Continental Italy

and France constitute anintermediate group, which displays equilibrium

be-tweenwesternandeasterncontributionstothegenepool. Theindividualvalues

oftheEast-West ratioforthe121cultivars aregiveninTableI.

3.3. Distribution of East-West genetic proximities of cultivars

withintheir country

There is a noticeabledispersion of the cultivars around thecorresponding

countrymeanfortheEast-Westratio;thisfactisillustratedbytheoverlapping

groups in the Newman-Keuls test of Table III. On the average, the

within-countryvariancescouldbeconsideredashomogeneous: Bartlett'shomogeneity

testwasonly9.44for10d.f. (probabilityofgreaterchi-squarewas0.51). This

validated our aim to draw general rules about the between-country pattern

of variability for this parameter. Moreover, the distribution of deviations of

-0,4

-0,3

-0,2

-0,1

0

0,1

0,2

0,3

0,4

-1

-0,8

-0,6

-0,4

-0,2

0

0,2

0,4

0,6

0,8

1

Axis 1 (83.43% of the discrimination)

A

xis

2

(13.44%

of

th

e

discrimination)

Corsica

Sicily

Continental France

Spain

Morocco

Algeria

Tunisia

Turkey

Syria

Israel

Greece

Continental Italy

Lebanon

Egypt

Yugoslavia

Portugal

A

0

2

4

6

8

10

12

14

16

Class interval

[-1

.947

/-1.

692

[

[-1

.692

/-1.

436

[

[-1

.436

/-1.

18

[

[-1

.1

8

/-0

.9

25

[

[-0

.925

/-0.

669

[

[-0

.669

/-0.

414

[

[-0

.414

/-0.

158

[

[-0

.158

/0

.097

[

[0

.097

/0

.353

[

[0

.353

/0

.608

[

[0

.608

/0

.864

[

[0

.864

/1

.1

2[

[1

.1

2

/1.

375

[

[1

.375

/1

.631

[

[1

.631

/1

.886

[

[1

.886

/2

.142

[

[2

.142

/2

.397

[

[

2.

397

/2

.653

[

[

2.

653

/2

.909

[

[2

.909

/3

.164

[

Number

of

ob

servat

io

ns

B

Figure2. Illustrationofthemeasurementofthegeneticproximitybetweenthe

cul-tivarsandtheeasternorwesternnaturalpopulations. A.Clusteringof121cultivars

from16countriesorregionsaccording totheirgenetic proximitytothecentroidsof

easternorwesternnaturalpopulations. Thecentroidcorrespondingtoagivenregion

isrepresentedbyalargersymbolthantheelementarycultivars. B.Distributionofthe

TableIII. Comparisonbetweenthestructureofcountriesfortheproximityoftheir

cultivarstooleastergenepoolsusing aDiscriminantAnalysis(A)andthesynthetic

parametercalledEast-Westratio (B).ThesignicancelevelofNewman-Keuls tests

forcomparisonofmeansis5%. Thecorrelationmeasuresbetweenthetwoparameters

aregiven.

A.Distributionof thegroups ofcultivarsonthe discriminant axis1

Rank Country Axis1 Standard error

1 Corsica 0.712 0.138 2 Algeria 0.266 0.124 | 3 Spain 0.106 0.088 || 4 C.Italy 0.000 0.083 ||| 5 Tunisia

−

0.017 0.124 ||| 6 C.France

−

0.115 0.042 ||| 7 Israel

−

0.206 0.113 || 8 Greece

−

0.244 0.113 || 9 Syria

−

0.353 0.074 | 10 Turkey

−

0.382 0.138 | 11 Sicily

−

0.400 0.105 |

Averagestandarderror(harmonicmean): 0.108

B.Distribution ofthegroups ofcultivarsaccording to theE-W ratio

Rank Country E-Wratio Standard error

1 Corsica 69.155 3.572 2 Algeria 59.126 3.195 | 3 Spain 54.703 2.259 || 4 Tunisia 51.776 3.195 ||| 5 C.Italy 51.774 2.154 ||| 6 C.France 48.996 1.077 ||| 7 Israel 46.823 2.916 || 8 Greece 45.515 2.916 || 9 Turkey 43.276 3.572 || 10 Syria 43.119 1.909 || 11 Sicily 40.971 2.700 |

Averagestandarderror(harmonicmean): 2.776

in viewof oursample size (insignicantcoecientsof skewness and kurtosis,

0.34 and 3.08for 116values, respectively). This distribution is displayed in

Figure2B.Onthebasisofthestandardiseddeviationsfromthecountrymean

compared to aStudent's t distribution with 95 d.f., only twocultivars could

be considered as having a marginal RAPD prole: Koroneiki from Crete

(Greece) and Manzanilla from Spain n

◦

4 and n

◦

expectedbychance inasampleof116trees. ItisalsopossiblethatKoroneiki

wasintroduced intoGreecefromawesternarea,such asCyrenaique(Libya).

We can then conclude that everything happens as if the proximity of a

cultivar from the eastern and western gene pools could be predicted by its

countryorregionoforigin,theaveragedeviationsfromthesepredictionsbeing

the same for every country or region. This is consistent with a process of

domesticationwherethechoiceofthebreedingmaterialwasprimarilylimited

bythepossibilitiesofexchangeandthereforebythegeographicdistance. The

geneticimplicationsofthisconclusionwillbediscussedfurther.

3.4. Hybrid statusof westerncultivars

Theresultsdescribedabovedonotdirectlytellanythingaboutthecausesof

thegeneticproximitiesbetweenthecultivarsandthewesternoreasternnatural

populationsofolives. Inparticular,takingintoaccountthegeneralmechanism

of varietal diusion from East to West, many cultivars originating from this

second zone may be suspected to derivefrom hybrids betweentrees from the

twomain groups.

4. DISCUSSION

Itistheoreticallyclearthattheclassicationofclonesismuchmorecomplex

than a simple dispatching betweenonly two east and west groups, since the

present varieties resulted from a long series of migration, hybridisation and

introgression processes, which are not easy to unravel. Indeed, most of the

stepsthatledtothepresentgeneticstructureareunknown. Individualvalues

of theEast-Westratio ofcultivars presentedin TableI will provide themain

elementsforaninferenceabouttheirhybridstatus.

Thegeographicgeneticstructuresobservedwithmitotypes[4]andwith

nu-clear RAPDs, which correspond to a clear distinction between eastern and

western populations, likely reect the existence of the last refugial zones as

shownforother foresttrees, insects andanimals [27]. Thus, fromthese data,

atleasttwoMediterraneanrefugialzoneswereinferred: oneintheEastand

an-otherintheWest(Fig.3I).Naturalbarrierstogeneows(AdriaticSea,Libyan

and EgyptianDeserts) havecontributedto preservethis structure (Fig. 3II).

Palynologicaland anthracologicdatasupport the existenceof O. europaeain

theNearEastduringthelastglaciation[2,14,30]. Thereisalsoevidenceofsuch

apresenceinthewesternMediterraneanduringtheUpperPleistocene[22,31],

butthenumberofpalynologicalstudiesconsideringthelastglaciation(18,000

12,000BP)isstilllimited. Incontrast,numerousstudiesaboutthelast10,000

yearshaveshowntheexpansionoftheolivetreewithhumanactivityfromthe

I.

Refugial zone(s) in Maghreb

and Southern Europe

Refugial zone(s)

in the Near-East

?

II.

_{Geographical barriers to}

the natural diffusion:

Adriatic Sea and Libyan Desert

Human exploitation

of "wild" olive

Human exploitation

of "wild" olive

2,800 BP

4,500 - 3,500 BP

3,000 BP

3,000 BP

Main center of olive

tree domestication

5,700 - 5,500 BP

Introduction of Eastern Olive in

Western Mediterranean followed

by a multilocal selection of cultivars

III.

3,200 BP

4,500 BP

3,200 BP

Figure3. Scenarioproposedforexplainingdiusionofolivetreeinsidethe

Mediter-ranean Basin since the last glaciation. I.Persistence of refugialzones inthe West

andEastoftheMediterraneanBasin duringthelastglaciation(20,00012,000 BP);

II.Naturaldiusionofoleasterpopulationwithhumaninuence(12,0006,000 BP);

III. Domesticationof olive tree inthe Near-East and beginning of the diusion of

cultivated olives towards the Mediterranean Basin (since 5,500 BP). Approximate

datesof introductionare reportedfrom Terral[28]. : Selection of cultivarsfrom

Thecultivarclassicationestablishedon thebasisof RAPDswasnot

con-gruentwith their cytoplasmic classication. On the basisof the distribution

of mitotypesin wild andcultivated olives,weclearlyshowsthat thediusion

of more or less selected varieties was oriented from East to West.

Further-more, the genetic background of each cultivar was considered in relation to

the genetic diversity and geographical origin of the oleaster. This approach

providesobjectiveargumentsforclassifyingthecultivars. Wehaveprovedthat

theRAPDproleofcultivarsoriginatingfromalimitedgeographicarea

(coun-try orregion)wasnot independentof the genotypesof the wild surrounding

populations. Of course,this associationisnotverytightbut itisnonetheless

signicant. Wehavealsoshownthatsomeregionsorcountriescouldelaborate

their own varieties probablyby using onlytheir local genetic resources. The

mosttypicalexampleis givenbyCorsica,whichproduced mostlyoriginal

va-rieties (Sabina, Zinzala, Capanacce). Themodel that we propose to connect

theRAPDproleofcultivarsorgroupsofcultivarstotheEasternorWestern

groupsofoleastersisofcoursesimplied. Thesamplingofnaturalpopulations

used in thisstudy didnotallowtheuseofamorediversiedset ofwild

pop-ulations. Butthismodel seemsrealistic,bothbecausethere isagreat genetic

diversitybetweenthetwogroupsofEastern andWestern wildolives(Fig. 1),

and also becausewehaveshown that the East-West direction wasprivileged

forthevarietytransfers.

It issurprisingthat theclassicationofcultivarsfor theirproximityto the

western and eastern gene pools could be completely achieved by only using

the East-West ratio. This eciency of a simple model (only oneparameter)

meansthattheabsolutegeneticdistancesbetweencultivatedandoleastersadd

nothingto apredictionofthe originofthegene poolpossessedbytheformer

category only based on the relative distances. The genetic interpretation is

that,onaverage,thepointswhichcorrespondtothe11coordinatesmeasuring

the RAPD prole of the cultivars are not far from the straight line joining

the centroidsof thewestern and easternoleasters. Inother words,

t

is small

compared to

r

and

s

. Theresult is that the bias on

D

asan estimate of

D

0

is limited. Moreover,if

t = 0

,

r

and

s

areonthesamestraightlineand their

absolutevaluesaddnoinformation totheirratio(which mayinterpretedasa

simplechange ofunits: r andsare normalisedbytheirsum,

r + s

). Thelast

argumentisgiveninTableIwhereitisshownthattheEast-Westratioranges

from33.10(Zaituna,code31)to71.51(Sabina,code79),aswithahighvalue

of

t

, all the values of the parameter should be close to 50. Therefore, when

the valueof

D

is really close to 50, asformany ItalianorFrench clones (for

instance, Pendolino, code21, Moraiolo, code 28,Picholine, code 39, Olivière,

code42),this maybeastrongindicationforahybridoriginofsuchcultivars

between eastern and western natural populations. When moredata become

available onvariabilityof wildpopulations,itwill beinterestingto testother

possiblemodels,withanindividualisationofmorethantwogroupsofoleasters,

5. CONCLUSION: AMODEL FOROLIVE DOMESTICATION

Asthenumbersandthesizesofoursamplednaturalpopulationswererather

limited, we restrictedour conclusions to clear general results. First, the

do-mestication of Olea europaea in the Mediterranean Basin was a

disymmetri-cal process where the prevailing direction of exchanges of breeding material

wasfrom the Near-East towards the occidental countries of Europe and the

Maghreb (Fig. 3II). This general pattern of gene ow is clearly established

bythecontrastbetweenthegeneraldiusion ofthemitotypesME1 andME2

in thecultivars from westerncountriesand theabsence ofasymmetrical

dis-seminationofthemitotypesMOMand MCKin thevarietiesproducedin the

eastern Mediterranean. Such a process of old variety transfers from eastern

Mediterranean Basincountries isalso in agreementwith historicdata, which

indicate that olivebreeding began in the Near-East[33]. Theprivileged

ma-ternal lineageamongthecultivars,as tracedbytheirmitotype,is alsological

because, before the twentieth century, the only practical possibility of

long-distancetransferwaslimitedtotheseedsandthevegetativepropagules.

Thesecondclearconclusionisthat theancientandlong-lastingdiusionof

moreorlessimprovedmaterialhasnotresultedinageneralmix ofcultivated

genotypes. Onthecontrary,thepresentvarietieshavekeptamemory ofthe

nuclear genotypesof the surrounding populations of oleasters. The practical

consequenceofthismemoryisthepersistenceofastructureforcultivars. This

is in accordance with the genetic resources of their country of origin. This

meansthat long-distancevarietytransportswerelikelyassociatedwith

multi-plecentresof domesticationusing thelocally availablenaturalpopulationsto

producecultivarsadaptedtothespecicculturalconditions,ecological

limita-tionsand traditionalhabits(Fig. 3III). Butmoreprosaicreasonsto uselocal

oleasterscouldhavebeenareducedcostandalsothediculty ofmultiplying

quicklyandgreatlyextenttheintroducedvarieties.

As thestructure ofthe RAPDprolesgaveclearevidence of theexistence

oftwomain groupsof oleasters(Eastand Westgroups), webuiltasimplied

modeltoconnectthenucleargenotypeofthecultivarstothesetwoclustersof

oleasters. Duetothelimitedanalysedsamples,amoresophisticatedmodelling

wouldbestatisticallyinecient. However,althoughsimplied,themodeloers

anecientwaytoprovidearoughevaluationofthedegreeofrelatednessof

in-dividualcultivarsorgroupsofcultivars(accordingtothecountryorregion,the

geographicareaortosomeotherclusteringmethod)withtwomaincontrasted

genepools. Inthefuture,amoreintensivesamplingofMediterraneanoleasters

and othergenemarkersas microsatellites willallowamoreprecise

character-isation of the relatedness of cultivars with a series of well-dened groups of

ACKNOWLEDGEMENTS

TheBRGand theFAIRprogramme CT95-0689supported thiswork. We

thankallourcorrespondentswhosenttousthesamplesofolivetreesanalysed

inthisstudy: DrG.Abdullah,DrM.Amirouch,DrA.H.Arsel,Prof. R.Assaf,

DrL.Baldoni,DrN.Gamoudi,DrA.Moukhli,DrA.Ouksili,DrL.Rallo,Dr

P. Vargas. We thank particularly Prof. P Villemur, and Prof. F. Dosbafor

theirsupport.

REFERENCES

[1] BaradatP.,LabbéT.,OPEP:Unlogicielintégrépourl'améliorationdesplantes

pérennes,in: Traitementsstatistiquesdesessaisdesélection.Stratégiesde

sélec-tiondesplantespérennes,CIRAD-CP(Ed.),Montpellier,1995,pp.303330.

[2] BaruchU.,BottemaS.,PalynologicalevidenceforclimaticchangesintheLevant

ca.17,000-9,000B.P.,inTheNatuanCultureintheLevant,Bar-YoseyO.,Vall

F.(Eds.),InternationalMonographsinPrehistory,AnnArbor,1991,pp.1120.

[3] BesnardG., KhadariB., VillemurP.,Bervillé A.,Cytoplasmicmalesterilityin

theolive(OleaeuropaeaL.),Theor.Appl.Genet.100(2000)10181024.

[4] BesnardG.,BervilléA.,MultipleoriginoftheMediterraneanolivededucedfrom

mitochondrialDNApolymorphism.C.R.Acad.Sci. III323(2000)173181.

[5] Besnard G., Baradat P., Bervillé A., Genetic relationships in the olive (Olea

europaea L.) reect multilocal selection of cultivars, Theor. Appl. Genet. 102

(2001)inpress.

[6] BesnardG.,BaradatP.,ChevalierD.,TagmountA.,BervilléA.,Genetic

dier-entiationintheolive (Oleaeuropaea) complexrevealedbyRAPDsand RFLPs

intherRNAgenes,Genet.Res.CropEvol.48(2001)inpress.

[7] BrunA.,MicrooresetPaléovégétations enAfriqueduNorddepuis30000ans,

Bull.Soc.Géol.France1(1989)2533.

[8] CarrionJ.S.,DupréM.,FumanalM.P.,MontesR.,Apalaeoenvironmentalstudy

in semi-arid southeastern Spain: The palynological and sedimentological

se-quenceatPernerascave(Lorca,Murcia),J.Archaeol.Sci. 22(1995)355367.

[9] DurhamR.E.,KorbanS.S.,EvidenceofgeneintrogressioninappleusingRAPD

markers,Euphytica79(1994)109114.

[10] Elant H.,Olive,inEvolution ofcrop plants,M.W. SimmondsEd.,Edinburgh,

Scotland,1976,pp.219221.

[11] Follieri M., Magri D.,Sadori L., 250,000-year pollen recordfromValledi

Cas-tiglione(Roma),PollenSpores30(1988)329356.

[12] KatoS., Ishikawa S., Imakawa S., Komori S., Mikami T., ShimamotoY.,

Cy-toplasmic relatedness of apple landraces and cultivars: a molecular analysis,

Euphytica66(1993)99102.

[13] KellyM.G.,HuntleyB., An11000-year recordofvegetation andenvironment

fromLagodiMartignano,Latium,Italy,J.Quat.Sci.6(1991)209224.

[14] KislevM.E.,NadelD.,CarmiI.,Epipalaeolithic(19,000BP)cerealandfruitdiet

[15] LashermesP.,AndrzejewskiS.,BertrandB.,CombesM.C.,DussertS.,Graziosi

G., Trouslot, P., Anthony F., Molecular analysis of introgressive breeding in

coee(CoeaarabicaL.),Theor.Appl.Genet.100(2000)139146.

[16] LebartL.,MorineauA.,PironM.,Statistiqueexploratoiremultidimensionnelle,

2ndedition,Paris,Dunod,1997.

[17] LerceteauE., RobertT., PétiardV.,CrouzillatD., Evaluationoftheextentof

geneticvariability amongTheobroma cacao accessionsusingRAPDandRFLP

markers,Theor.Appl.Genet.95(1997)1019.

[18] LumaretR.,OuazzaniN.,MichaudH.,VillemurP.,Cultivatedoliveandoleaster:

twoverycloselyconnectedpartnersofthesamespecies(Oleaeuropaea).Evidence

fromenzymepolymorphism,Bocconea7(1997)3942.

[19] MagriD.,LateQuaternaryvegetationhistoryatLagaccionenearLagidiBolsena

(centralItaly),Rev.Palaeobot.Palynol.106(1999)171208.

[20] MagriD.,SadoriL.,LatePleistoceneandHolocenepollenstratigraphyatLago

diVico(centralItaly),Veg.Hist.Archaeobot.8(1999)247260.

[21] Pérez-ObiolR.,JuliàR.,ClimatechangeontheIberianPeninsularecordedina

30,000-YrpollenrecordfromLakeBanyoles,Quat.Res.41(1994)9198.

[22] Pons A., Reille M., The Holocene- and upper Pleistocene pollen record from

Padul(Granada, Spain): anewstudy,Palaeogeogr.Palaeoclimatol.Palaeoecol.

66(1988)243263.

[23] Quillet M.C., Madjidian N., Griveau Y., Serieys H., Tersac M., Lorieux M.,

BervilléA., Mapping of genetic factors controllingpollenviabilityin an

inter-speciccrossinHelianthussect.Helianthus,Theor.Appl.Genet.91(1995)1195

1202.

[24] Riera-MoraS.,Esteban-AmatA.,Vegetationhistoryandhumanactivityduring

thelast6000yearsonthecentralCatalancoast(northeasternIberianPeninsula),

Veget.Hist.Archaeobot.3(1994)723.

[25] Saporta G., Probabilité, Analyse des Données et Statistique, Paris, Technip,

1990.

[26] SolariM.E., VernetJ.L.,Lateglacialand HolocenevegetationoftheCorbières

basedoncharcoalanalysisattheCovadeL'Espérit(Salses,PyrénéesOrientales,

France),Rev.Paleobot.Palynol.71(1992)111120.

[27] Taberlet P.,Fumagalli L., Wust-SaucyA.G., CossonJ.F., Comparative

phylo-geography and postglacial colonization routes in Europe, Mol. Ecol. 7 (1998)

453464.

[28] Terral J.F., La domestication de l'olivier (Olea europaea L.) en Méditerranée

nord-occidentale : Approchemorphométriqueetimplicationspaléoclimatiques,

Thèse,UniversitéMontpellierII,France,1997.

[29] TerralJ.F., Arnold-SimardG., Beginningsofolivecultivationineastern Spain

inrelationtoHolocenebioclimaticchanges,Quat.Res.46(1996)176185.

[30] VanZeist W., Woldring H., Holocene vegetation and climateof Northwestern

Syria,Palaeohistoria22(1980)111125.

[31] WattsW.A.,AllenJ.R.M.,HuntleyB.,Vegetationhistoryandpalaeoclimateof

thelastglacialperiodatLagoGrandediMonticchio,SouthernItaly,Quat.Sci.

Rev.15(1996)133153.

[32] YllE.I.,Perez-ObiolR.,Pantaleon-CanoJ.,RoureJ.M.,Palynologicalevidence

for climaticchangeand humanactivity during Holocene onMinorca (Balearic

[33] ZoharyD., Hopf M., Domestication of plants inthe Old World, 2nd edition,

OxfordClarendonPress, 1994,pp.137142.

[34] ZoharyD.,Spiegel-RoyP.,BeginningsoffruitgrowingintheOldWorld,Science