Molecular Ecology (2003)

12

1451ndash1460 doi 101046j1365-294X200301810x

copy 2003 Blackwell Publishing Ltd

Blackwell Publishing Ltd

Chloroplast DNA phylogeography reveals colonization history of a Neotropical tree

Cedrela odorata

L in Mesoamerica

S CAVERS

C NAVARRO

dagger

and A J LOWE

Centre for Ecology and Hydrology-Edinburgh Bush Estate Penicuik Midlothian EH26 0QB UK

dagger

Centro Agroacutenomico Tropical de Investigacioacuten y Ensentildeanza Cartago Turrialba 7170 Costa Rica

Abstract

Spanish Cedar (

Cedrela odorata

L) is a globally important timber species which has beenseverely exploited in Mesoamerica for over 200 years Using polymerase chain reactionndashrestriction fragment length polymorphisms its chloroplast (cp) DNA phylogeography wasstudied in Mesoamerica with samples from 29 populations in six countries Five haplotypeswere characterized phylogenetically grouped into three lineages (Northern Central andSouthern) Spatial analysis of ordered genetic distance confirmed deviation from a patternof isolation by distance The geographically proximate Northern and Central cpDNAlineages were genetically the most differentiated with the Southern lineage appearingbetween them on a minimum spanning tree However populations possessing Southernlineage haplotypes occupy distinct moist habitats in contrast to populations possessingNorthern and Central lineage haplotypes which occupy drier and more seasonal habitatsGiven the known colonization of the proto-Mesoamerican peninsula by South Americanflora and fauna prior to the formation of the Isthmus of Panama it seems most likely thatthe observed population structure in

C odorata

results from repeated colonization ofMesoamerica from South American source populations Such a model would imply anancient pre-Isthmian colonization of a dry-adapted type (possessing the Northern lineageor a prototype thereof) with a secondary colonization via the land bridge Following thisa more recent (possibly post-Pleistocene) expansion of moist-adapted types possessing theSouthern lineage from the south fits the known vegetation history of the region

Keywords

Cedrela odorata

differentiation dispersal Meliaceae Spanish Cedar universal cpDNAmarkers

Received 19 December 2002 revision received 13 January 2003 accepted 13 January 2003

Introduction

Phylogeography examines the correspondence betweengenetic relationships and geographical distribution (Avise

et al

1987) Population genetic structure is as much aproduct of history as of present-day migration patternsand isolation of populations hence a synthesis of gene-alogical data with independent information includinggeology palynology and archaeology (Avise

et al

1987Bermingham amp Moritz 1998) may disentangle the histor-ical component of population structure from that which isthe result of contemporary gene flow processes

In plants the recent development of universal primersets targeting noncoding regions of the chloroplast(cp) genome (Taberlet

et al

1991 Demesure

et al

1995Dumolin-Lapegue

et al

1997b Hamilton 1999) has revealedsubstantial amounts of intraspecific variation (Newton

et al

1999a) and cpDNA data have now been successfully usedfor several phylogeographic studies of plants (Petit

et al

1993 Petit

et al

1997 Caron

et al

2000 Dutech

et al

2000Raspe

et al

2000) As a result of its usual maternal inherit-ance in angiosperms cpDNA is transmitted only throughseeds and therefore has less potential for gene flow thannuclear genes which can also move by pollen dispersalConsequently genetic variation in the chloroplast genomeis often more highly geographically structured than thatin the nuclear genome Furthermore as the rate of cpDNA

Correspondence S Cavers Fax 0131 4453943 E-mail scavcehacuk

1452

S C A V E R S C N A V A R R O and A J L O W E

copy 2003 Blackwell Publishing Ltd

Molecular Ecology

12 1451ndash1460

sequence evolution is slow (Wolfe

et al

1987) observedpatterns reflect the outcome of processes over long time-scales (Ennos

et al

1999) so cpDNA is ideal for studyinghistorical patterns of gene flow in particular migrationand colonization

Several recent studies have taken advantage of thesecharacteristics to investigate vegetation changes in par-ticular those related to glacial cycles In Europe (

Alnus glu-tinosa

King amp Ferris 1998

Quercus

sp Dumolin-Lapegue

et al

1997a) North America (

Dryas integrifolia

Tremblayamp Schoen 1999

Liriodendron tulipifera

Sewell

et al

1996)and the tropics (

Vouacapoua americana

Dutech

et al

2000

Dicorynia guianensis

Caron

et al

2000) cpDNA has beensuccessfully used to detect spatiotemporal patterns offragmentation and dispersal resulting from climatic vari-ations during the Pleistocene epoch

The region of interest in this study is tropical Mesoamericabetween southern Mexico and northern Colombia encom-passing Guatemala Honduras Nicaragua Costa Ricaand Panama Present day Mesoamerica is a region richin diversity there are still more than a quarter of a millionsquare kilometres of primary vegetation around 24 000plant species (of which some 5000 are endemic) and nearly3000 vertebrate species (of which over 1000 are endemicMyers

et al

2000)The distribution and composition of the Mesoamerican

flora and fauna have been strongly influenced by geo-logical and climatic events (Burnham amp Graham 1999)Prior to the formation of the Isthmus of Panama there wasconsiderable interchange of flora between the separateland masses of North and South America (Raven amp Axelrod1974) possibly via an island chain At this time populationswould have been isolated in the proto-Mesoamericanpeninsula by the sea to the south and by the more tem-perate climate to the north (Savage 1982) Following theformation of the Panamanian land link [between 5and 3 million years ago (Ma) Coney 1982] the GreatAmerican Interchange resulted in numerous invasionsof Mesoamerica by South American angiosperm flora(Burnham amp Graham 1999) Later the climatic fluctu-ations of the Pleistocene (16ndash001 Ma) had a substantialinfluence on the Mesoamerican flora (Prance 1982abToledo 1982) Major fragmentation of the extensive tro-pical forest took place (Toledo 1982 Leyden 1984 Islebe ampHooghiemstra 1997 Williams

et al

1998 Hewitt 2000)with many species restricted to refuge populations in theregion of present day Guatemala and northwest Colombiaduring glacial maxima In general for many plant andanimal species of Mesoamerica distinct biogeographicpatterns remain reflecting the significant influence of dis-persal and isolation extinction and colonization on thepopulations of this dynamic and diverse region (Savage1982 Bermingham amp Martin 1998 Burnham amp Graham1999)

Spanish Cedar (

Cedrela odorata

) is a neotropical memberof the hardwood family Meliaceae (Swietenioideae) wellknown for its high-quality high-value timber The species(and family) has a long history of human exploitation andis still a valuable commodity used in furniture-makingand construction (Lamb 1968 Rodan

et al

1992 Valera1997)

Cedrela odorata

is naturally distributed from theMexican Pacific coast at 26

deg

N and the Mexican Atlanticcoast at 24

deg

N throughout the Caribbean islands the Yuca-tan and lowland Central and South America to northernArgentina at 28

deg

S The tree is deciduous and grows in bothdry and moist lowland areas where soils are not floodedup to around 1200 m altitude It is a fast-growing light-demanding species (Lamb 1968 Chaplin 1980 Valera1997) reaching 40 m in height and 120 cm in diameter Aswith other Meliaceae the species is monoecious (flowersare unisexual Pennington

et al

1981 pollinated by smallbees wasps and moths Navarro

et al

2002 Bawa

et al

1985) Flowering occurs annually in 10ndash15-year-old treesand a good seed crop is produced every 1ndash2 years (ICRAFAgroforestree database httpwwwicrafcgiarorg) Seedis wind dispersed

The value of

C odorata

has resulted in over-exploitationof the species in its natural habitat for two centuries Anumber of studies have been carried out on the specieswithin Costa Rica focusing on its known intraspecificvariation manifest primarily as tolerance for both dryand moist habitats

Common garden experiments haveindicated that this environmental lsquotolerancersquo has a gen-etic basis and that ecotypic differentiation has occurredwithin the species For example apical dominance experi-ments (Newton

et al

1995)

Hypsipylla grandella

resistancetrials (Newton

et al

1999b) and morphological studies(Navarro

et al

2002) all identify two distinct groups withinCosta Rica that are correlated with habitat Randomamplified polymorphic DNA (RAPD) analysis of CostaRican populations (Gillies

et al

1997) showed differenti-ation between these ecotypes for neutral loci To datethere has been no large-scale molecular study of

C odorata

and the extent of intraspecific variation in the Meso-american population is unknown Given the threat posedto the species by unsustainable logging practices andhabitat loss and evidence indicating significant popula-tion structuring there is a clear need to assess the currentlevels and distribution of diversity in the Mesoamericanpopulation

This study investigates the phylogeographic structureof the Mesoamerican population of

C odorata

Universalchloroplast markers are employed to identify patternsof population structure that reflect the seed dispersalhistory of the species assuming maternal inheritance of thechloroplast The results are interpreted in the light of theknown geological and climatic history of the MesoamericanIsthmus to infer colonization dynamics of

C odorata

P H Y L O G E O G R A P H Y O F

C E D R E L A O D O R A T A

1453

copy 2003 Blackwell Publishing Ltd

Molecular Ecology

12 1451ndash1460

Materials and methods

Samples were collected from a total of 580

Cedrela odorata

individuals in 29 populations throughout Mesoamerica(Fig 1 and Table 1) Populations were defined as groups oftrees at least 100 m apart but within a coherent geographicalarea such that they were in potential reproductive contactTwenty trees were sampled per population Four populationswere sampled from each of Mexico Guatemala Hondurasand Nicaragua Three populations were sampled fromPanama and 10 from Costa Rica Individuals were sampledby collecting either leaf tissue (dried on silica gel) or cam-bium tissue [immersed in 70 30 ethanol cetyltrime-thylammonium bromide (CTAB) buffer containing 100 m

m

TrisndashHCl pH 80 20 m

m

ethylenediaminetetraacetic acid(EDTA) 14

m

NaCl 1 polyvinylpyrrolidone-40T (PVPndash40T) 2 CTAB] Genomic DNA was extracted using amodified CTAB protocol (Gillies

et al

1997)Screening for variation in the cpDNA used the universal

primers described in Demesure

et al

(1995) Dumolin-Lapegue

et al

(1997b) and Hamilton (1999) The poly-merase chain reaction (PCR) protocol was as describedin Demesure

et al

(1995) and fragment patterns were visu-alized on 8 nondenaturing polyacrylamide gel in a Hoefer

SE600 electrophoresis unit (300 V) using Trisndashborate EDTAbuffer (1

times

) As a result of amplification problems not allpopulations could be screened for all 20 individuals col-lected and those from Guatemala and Panama wereassessed using only five individuals Pons amp Petit (19951996) emphasize the importance of increased numbers ofpopulations as opposed to increased within-populationsample sizes for studies of population subdivision so thereduced data sets from Guatemala and Panama wereincluded in the analysis

All individuals were characterized for cpDNA haplo-type and the data set was analysed for within-population(

H

S

) and total (

H

T

) diversity and for the level of popula-tion subdivision (

G

ST

) A minimum spanning tree (Excoffieramp Smouse 1994) was constructed using

minspnet

(avail-able at httplgbunigechsoftware) scoring fragmentsas multistate characters then preparing a pairwise dis-tance matrix based on the number of mutational (indelor restriction site) differences between haplotypes Byincorporating the relationships between haplotypes fromthe minimum spanning tree an estimate of populationsubdivision for phylogenetically ordered alleles (

N

ST

) wasobtained and a test statistic

U

comparing the values of

N

ST

and

G

ST

was calculated (Pons amp Petit 1996) All statisticswere calculated using the program

haplonst

which isavailable at httpwwwpierrotoninrafrgeneticslaboSoftware)

The geographical component of population structurewas investigated through spatial analysis of genetic variationPairwise values of

F

ST

(Wright 1969) unordered geneticdistance

D

G

(Gregorius 1978 where

and

i

j

are two populations

n

is the number of haplotypesand

p

ik

is the frequency of the

i

th haplotype in the

k

thpopulation) and ordered genetic distance (taking the dis-tance between haplotypes as the number of mutational stepsbetween them along the minimum spanning networkan average distance between populations was then deter-mined based on this distance and the frequency of eachhaplotype in the populations) were plotted against geogra-phical distances The plots of

F

ST

and unordered geneticdistance were prepared using the program

sgs

(Degen

et al

2001 available at httpkourouciradfrgenetiquesoftwarehtml)

Finally environmental data for each of the populationsites (mean annual precipitation mean number of drymonths and mean annual temperature Table 1) wereinvestigated with sites grouped according to their popula-tion cpDNA haplotype A one-way analysis of variance(

anova

using

minitab

) was carried out to look for associ-ations between the environmental data and cpDNA haplo-type group

Fig 1 Map of the populations and distribution of haplotypes ofCedrela odorata sampled in Mesoamerica Inset shows the locationof Mesoamerica on a global map Population markers areproportional to the number of samples analysed large circles 20individuals small circles five individuals

D i j p pG ik jkk

n

( ) = minus=

sum12 1

1454

S C A V E R S C N A V A R R O and A J L O W E

copy 2003 Blackwell Publishing Ltd

Molecular Ecology

12 1451ndash1460

Results

Six polymorphic primerenzyme combinations wereidentified (Table 2) These contained a total of 11 indels(insertiondeletion mutations) and one restriction sitemutation characterizing five cpDNA haplotypes (Table 2)Half of the mutations differentiated Mexican and Guate-malan populations from all other populations The fivehaplotypes could be clearly resolved using just three primerenzyme combinations and these alone were used to screenthe whole collection

Almost all of the populations were fixed for a single haplo-type (Table 3) hence average within-population diver-sity was very low

H

S

= 003 (Table 4) Only three of the 29

populations screened showed any within-populationdiversity San Francisco in Panama Upala in Costa Ricaand Los Esclavos in Guatemala Of these the first two areat boundaries between populations fixed for different haplo-types while Los Esclavos had the only case of a privatehaplotype detected during the study The total diversity(

H

T

) was 070 and the level of population subdivision (

G

ST

)was 096 (Table 4)

The minimum spanning network of haplotype relation-ships (Fig 2) shows three haplotype lineages The first is inMexico and Guatemala (including two haplotypes hence-forth called lsquoNorthernrsquo lineage) the second in HondurasNicaragua and northwestern Costa Rica (a single haplo-type lsquoCentralrsquo) and the third in eastern and southwestern

Table 1 Details of locations and environmental data for all populations sampled

PopulationLatitude (N)

Longitude (W)

Sampling method

Altitude (m)

No of dry months

Precipitation (mmyear)

Mean annual temperature (degC)

PanamaGualaca 8deg35prime 82deg14prime cambium 150 4 4000 269Las Lajas 8deg12prime 81deg52prime cambium 20 4 3500 269San Francisco 8deg14prime 80deg59prime cambium 100 4 2500 267

Costa RicaHorizontes 10deg44prime 85deg35prime leaf 200 5 1500 270Puriscal 09deg56prime 84deg17prime leaf 900 5 2400 250Canas 10deg12prime 84deg57prime leaf 100 5 1829 270Palo Verde 10deg21prime 85deg21prime leaf 50 5 1824 270Hojancha 10deg05prime 85deg22prime cambium 250 5 2232 270Jimenez 10deg13prime 83deg37prime leaf 240 1 4465 221Talamanca 09deg38prime 82deg51prime leaf 75 0 2467 256Upala 10deg47prime 85deg02prime cambium 150 4 2558 250Pacifico Sur 08deg32prime 82deg51prime cambium 40 0 4089 268Perez Zeledon 09deg20prime 83deg39prime cambium 700 4 2934 233

NicaraguaOmetepe 11deg29prime 85deg29prime leaf 40 6 1500 273Masatepe 11deg54prime 86deg08prime leaf 450 6 1880 273Wabule 12deg53prime 85deg41prime leaf 600 6 1750 273La Trinidad 12deg59prime 86deg14prime leaf 600 6 1700 273

HondurasMeambar 14deg50prime 88deg06prime leaf 600 1 2425 220Taulabe 14deg50prime 88deg06prime leaf 633 1 2425 220Comayagua 14deg09prime 87deg37prime leaf 579 5 1052 244La Paz 14deg25prime 87deg03prime leaf 726 3ndash4 1976 220

GuatemalaLos Esclavos 14deg15prime 90deg17prime leaf 737 6 1552 236El Idolo 14deg26prime 91deg23prime leaf 300 5 3010 260Tikal 17deg14prime 89deg37prime leaf 250 5 2500 300San Jose 17deg11prime 89deg52prime leaf 235 5 1300 260

MexicoCalakmul 18deg27prime 88deg19prime leaf 300 5 1450 245Bacalar 18deg17prime 89deg09prime leaf 15ndash300 5 1450 240Zona Maya 19deg22prime 88deg01prime leaf 50 5 1450 240Escarcega 18deg24prime 90deg54prime leaf 100 5 1198 240

P H Y L O G E O G R A P H Y O F C E D R E L A O D O R A T A 1455

copy 2003 Blackwell Publishing Ltd Molecular Ecology 12 1451ndash1460

Costa Rica and Panama (including two haplotypes lsquoSouth-ernrsquo) The network does not reflect geographical distributionas the Northern lineage was most strongly differentiatedfrom the geographically proximate Central lineage ratherthan from the geographically distant Southern lineage

The NST estimate (NST = 098) was greater than the GSTestimate (GST = 096 Table 4) although the difference wasnot significant Nevertheless the phylogenetic compon-ent of the geographical distribution was clearly evident inthe contrast between spatial analysis of ordered and un-ordered data (Fig 3) Pairwise FST remained fairly constantaround the overall GST of 096 (Fig 3 top panel) Howeverthe unordered genetic distance increased steadily withpairwise geographical distance reaching 1 for distanceclasses above 880 km (Fig 3 top panel) Unordered geneticdistance reflected only the number of haplotypes sharedbetween populations and did not take into account phylo-genetic relationships Hence it increased with distance ascomparisons between populations from different geo-graphical regions are unlikely to share any haplotypes Incontrast the ordered genetic distance increased with geo-graphical distance between populations up to around700 km but then fell above distances of 1000 km (Fig 3bottom panel) This is because of the large genetic dis-tance between the geographically proximate Northern andCentral lineages in comparison to the lesser divergencebetween the Northern and Southern lineages (when haplo-type relationships are taken into account Fig 2)

Investigation of environmental data showed significantdifferences between sites with populations possessing dif-ferent cpDNA lineages (Fig 4 population Upala wasclassed as being of the Southern lineage) for both meanannual rainfall (anova F = 1760 P lt 0000 df = 2) andnumber of dry months (anova F = 592 P lt 0008 df = 2)The analysis of variance for mean annual temperature wasnot significant Therefore populations possessing North-ern or Central lineage haplotypes were found to experience

generally low annual rainfall (means of 1738 and 1884 mmyear respectively) and pronounced seasonality (greaterannual number of dry months mean 404) Populationspossessing Southern lineage types occupied wetter siteswith shorter dry seasons (mean rainfall 3314 mmyearmean duration of dry season 26 months)

Discussion

The level of population differentiation within Cedrelaodorata (GST = 096) is one of the highest yet obtained (cfQuercus sp GST = 0828 Dumolin-Lapegue et al 1997aVouacapoua americana GST = 089 Dutech et al 2000 Arganiaspinosa GST = 060 El Mousadik amp Petit 1996) Consider-ing the geographical distances (maximum interpopulationdistance approx 1500 km) and environmental gradientsinvolved relative to seed and pollen dispersal distancesin the species (maximum distance approx hundredsof metres) the collection cannot be panmictic and somegenetic structuring was to be expected because of geneticdrift However several aspects of the haplotypic distribu-tion indicate divergence from a neutral pattern of isolation bydistance

First seed flow between populations possessing theNorthern and Central lineages appears to be restrictedThere is no clear ecological or physical reason why geneflow via seed should not occur yet the boundary is dis-tinct In European oaks the recolonization of northernEurope has occurred in parallel from three genetically dif-ferentiated refugial populations since the Pleistoceneestablishing a distinct pattern in cpDNA variation (Dumolin-Lapegue et al 1997a Petit et al 2002ab) At and beyondcontact zones between migrating fronts significant mixingof haplotypes has occurred as all populations are coloniz-ing new territory However following the establishment ofa population immigration is rare (most recruitment is fromlocal sources) and an enduring patch structure results For

Table 2 Description of the five cpDNA haplotypes identified

Polymorphic fragments

PF1 PF2 PF3 PF4 PF5 PF6 PF7 PF81ndash510 1ndash760 1ndash380 1ndash370 1ndash410 1ndash390 1ndash250 1ndash4152ndash240 2ndash740 2ndash360 2ndash340 2ndash400 2ndash330 2ndashabsent 2ndash4103ndash230 3ndash390 3ndash320

Haplotype indel indel indel indel indel indel site indel

1 2 1 1 1 1 2 1 12 2 1 1 1 3 1 1 13 1 2 2 2 1 3 2 24 3 2 1 2 2 3 2 25 3 2 1 2 2 2 2 2

Numbers 1 2 3 indicate character state of fragment in decreasing order of size (bp)Full details of primersenzymes available from authors

P H Y L O G E O G R A P H Y O F C E D R E L A O D O R A T A 1455

copy 2003 Blackwell Publishing Ltd Molecular Ecology 12 1451ndash1460

Costa Rica and Panama (including two haplotypes lsquoSouth-ernrsquo) The network does not reflect geographical distributionas the Northern lineage was most strongly differentiatedfrom the geographically proximate Central lineage ratherthan from the geographically distant Southern lineage

The NST estimate (NST = 098) was greater than the GSTestimate (GST = 096 Table 4) although the difference wasnot significant Nevertheless the phylogenetic compon-ent of the geographical distribution was clearly evident inthe contrast between spatial analysis of ordered and un-ordered data (Fig 3) Pairwise FST remained fairly constantaround the overall GST of 096 (Fig 3 top panel) Howeverthe unordered genetic distance increased steadily withpairwise geographical distance reaching 1 for distanceclasses above 880 km (Fig 3 top panel) Unordered geneticdistance reflected only the number of haplotypes sharedbetween populations and did not take into account phylo-genetic relationships Hence it increased with distance ascomparisons between populations from different geo-graphical regions are unlikely to share any haplotypes Incontrast the ordered genetic distance increased with geo-graphical distance between populations up to around700 km but then fell above distances of 1000 km (Fig 3bottom panel) This is because of the large genetic dis-tance between the geographically proximate Northern andCentral lineages in comparison to the lesser divergencebetween the Northern and Southern lineages (when haplo-type relationships are taken into account Fig 2)

Investigation of environmental data showed significantdifferences between sites with populations possessing dif-ferent cpDNA lineages (Fig 4 population Upala wasclassed as being of the Southern lineage) for both meanannual rainfall (anova F = 1760 P lt 0000 df = 2) andnumber of dry months (anova F = 592 P lt 0008 df = 2)The analysis of variance for mean annual temperature wasnot significant Therefore populations possessing North-ern or Central lineage haplotypes were found to experience

generally low annual rainfall (means of 1738 and 1884 mmyear respectively) and pronounced seasonality (greaterannual number of dry months mean 404) Populationspossessing Southern lineage types occupied wetter siteswith shorter dry seasons (mean rainfall 3314 mmyearmean duration of dry season 26 months)

Discussion

The level of population differentiation within Cedrelaodorata (GST = 096) is one of the highest yet obtained (cfQuercus sp GST = 0828 Dumolin-Lapegue et al 1997aVouacapoua americana GST = 089 Dutech et al 2000 Arganiaspinosa GST = 060 El Mousadik amp Petit 1996) Consider-ing the geographical distances (maximum interpopulationdistance approx 1500 km) and environmental gradientsinvolved relative to seed and pollen dispersal distancesin the species (maximum distance approx hundredsof metres) the collection cannot be panmictic and somegenetic structuring was to be expected because of geneticdrift However several aspects of the haplotypic distribu-tion indicate divergence from a neutral pattern of isolation bydistance

First seed flow between populations possessing theNorthern and Central lineages appears to be restrictedThere is no clear ecological or physical reason why geneflow via seed should not occur yet the boundary is dis-tinct In European oaks the recolonization of northernEurope has occurred in parallel from three genetically dif-ferentiated refugial populations since the Pleistoceneestablishing a distinct pattern in cpDNA variation (Dumolin-Lapegue et al 1997a Petit et al 2002ab) At and beyondcontact zones between migrating fronts significant mixingof haplotypes has occurred as all populations are coloniz-ing new territory However following the establishment ofa population immigration is rare (most recruitment is fromlocal sources) and an enduring patch structure results For

Table 2 Description of the five cpDNA haplotypes identified

Polymorphic fragments

PF1 PF2 PF3 PF4 PF5 PF6 PF7 PF81ndash510 1ndash760 1ndash380 1ndash370 1ndash410 1ndash390 1ndash250 1ndash4152ndash240 2ndash740 2ndash360 2ndash340 2ndash400 2ndash330 2ndashabsent 2ndash4103ndash230 3ndash390 3ndash320

Haplotype indel indel indel indel indel indel site indel

1 2 1 1 1 1 2 1 12 2 1 1 1 3 1 1 13 1 2 2 2 1 3 2 24 3 2 1 2 2 3 2 25 3 2 1 2 2 2 2 2

Numbers 1 2 3 indicate character state of fragment in decreasing order of size (bp)Full details of primersenzymes available from authors

1456 S C A V E R S C N A V A R R O and A J L O W E

copy 2003 Blackwell Publishing Ltd Molecular Ecology 12 1451ndash1460

C odorata the lack of cpDNA mixing at the contact zonesuggests that rather than parallel colonization secondarycontact has occurred between an established populationand a colonizing front

Biogeographical studies of the flora (Raven amp Axelrod1974) and fauna (Savage 1982) and fossil pollen data (Graham1999) indicate that there was considerable dispersal fromSouth America to Mesoamerica prior to the formation ofthe Panama land bridge (5ndash3 Ma Coney 1982) Given thelarge differentiation between the Northern and Central

lineages (suggesting a timescale in the order of millions ofyears King amp Ferris 1998 Ennos et al 1999 Dutech et al2000) it seems feasible to suggest that the Northern lineagecolonized the proto-Mesoamerican Isthmus prior to for-mation of the land bridge via long-distance dispersal ormigration across an early island arc and that the Centrallineage followed later probably across the land bridge (asoccurred for most lowland Mesoamerican flora Gentry1982a Burnham amp Graham 1999) Several other studieshave noted the distinctiveness of MexicanYucatan popu-lations from southern MesoamericanPacific coast popu-lations (eg Cordia alliodora Boshier 1984 Chase et al 1995Gliricidia sepium Lavin et al 1991 Calliandra calothyrsusChamberlain 1998 Swietenia macrophylla Gillies et al 1999)

Table 3 Distribution of haplotypes within populations of Cedrelaodorata in Mesoamerica

Haplotypes

Population 1 2 3 4 5 Total

PanamaGualaca 5 5Las Lajas 5 5San Francisco 1 4 5

Costa RicaHorizontes 20 20Puriscal 20 20Canas 20 20Palo Verde 20 20Hojancha 20 20Jimenez 20 20Talamanca 20 20Upala 19 1 20Pacifico Sur 20 20Perez Zeledon 20 20

NicaraguaOmetepe 20 20Masatepe 20 20Wabule 20 20La Trinidad 20 20

HondurasMeambar 20 20Taulabe 20 20Comayagua 20 20La Paz 20 20

GuatemalaLos Esclavos 4 1 5El Idolo 5 5Tikal 5 5San Jose 5 5

MexicoCalakmul 20 20Bacalar 20 20Zona Maya 20 20Escarcega 20 20

Total 99 1 259 106 9 475

Details of the haplotypes are given in Table 2

Table 4 Mean levels of population and total diversity and level ofpopulation subdivision for 29 populations of Cedrela odoratathroughout Mesoamerica

Number of populations 29Arithmetic mean population size 164Harmonic mean population size 1160HS 003 (002)HT 070 (005)GST 096 (003)νS 001 (001)νT 070 (004)NST 098 (001)U 141

Standard errors are given in brackets All estimates were calculated using the software haplonst NST is based on the minimum spanning tree in Fig 2

Fig 2 Minimum spanning tree for the five haplotypes identifiedin Cedrela odorata The geographical distribution of the haplotypesis shown in Fig 1 Bars on connecting spans indicate minimumnumbers of individual mutations (see Table 2 for details) a PF4(state 1ndashstate 3) b PF6 (1ndash2) c PF3 (1ndash2) d PF5 (1ndash2) e PF4 (1ndash2)f PF2 (1ndash2) g PF7 (1ndash2) h PF8 (1ndash2) i PF6 (2ndash3) j PF1 (1ndash2)k PF1 (1ndash3) l PF4 (1ndash2)

P H Y L O G E O G R A P H Y O F C E D R E L A O D O R A T A 1457

copy 2003 Blackwell Publishing Ltd Molecular Ecology 12 1451ndash1460

possibly indicating a general pattern which should be fur-ther investigated

In contrast the contact zone between the Central andSouthern lineages is quite different Trees possessing thetwo lineages occupy significantly different habitats interms of both annual rainfall and seasonality (Figs 1 and4 Table 1) Previous studies have demonstrated signi-ficant morphological (Navarro et al 2002) physiological(Newton et al 1995 1999b) and nuclear sequence (Gillieset al 1997) differentiation between populations that possessthe Central and Southern cpDNA lineage types indicating

probable reproductive isolation and the existence of eco-types On the basis of the current evidence the most likelyscenario for generation of the observed pattern is repeatedcolonization of the Isthmus during known fluctuationsin vegetation assemblages associated with the climaticchanges of the Pleistocene epoch (Graham 1991 Williamset al 1998 Hewitt 2000) Under such a model the Centrallineage could have colonized during a period thatfavoured expansion of dry-adapted vegetation followingthe formation of the Isthmus of Panama (see above) TheSouthern lineage would then represent a much morerecent northward colonization of a wet-adapted type froma previously differentiated southern American sourcepopulation during a period favouring expansion of moistforest assemblages This could have occurred recentlyfollowing the last glacial maximum (13 000 years ago)from a source in the geographically proximate wet Chocoregion of Colombia Interestingly within the Southernlineage two haplotypes are present in Panama whereasonly one is present in Costa Rica a pattern which mightbe expected from a northern expansion and associatedreduction in diversity in this region

Fig 3 Spatial analysis of genetic variation in Cedrela odoratacpDNA Top panel average pairwise comparisons of geneticdistance and FST by distance class (calculated using sgs softwareDegen et al 2001) Asterisk indicates significant positive correla-tion (at 95 confidence level) Bottom panel average pairwiseordered genetic distance by distance class based on the minimumspanning tree (Fig 2) All 95 confidence intervals less than 003In both plots the number of pairwise population comparisonsin each distance class is shown above the x-axis

Fig 4 Top panel group means and 95 confidence intervals forannual rainfall for all populations in the three principal haplotyperegions Bottom panel group means and 95 confidence intervalsfor number of dry months for all populations in the three principalhaplotype regions Results from the corresponding anova aregiven in the text

1458 S C A V E R S C N A V A R R O and A J L O W E

copy 2003 Blackwell Publishing Ltd Molecular Ecology 12 1451ndash1460

As is the case for many other Neotropical species biogeo-graphical evidence suggests that C odorata has colonizedMesoamerica from South America The species is distri-buted widely throughout the Neotropics but its range iscentred in South America Evidence for lsquowavesrsquo of colonizationsome from divergent South American sources has beennoted for other Central American species for examplefreshwater fish (Bermingham amp Martin 1998) In additionthe tropical moist forest was substantially and repeatedlyfragmented into refugial areas throughout Mesoamericaand northern South America (Gentry 1982b Prance1982ab Toledo 1982) Hence it seems likely that thehaplotype distribution pattern in Mesoamerica is causedby repeated northward colonization from previously dif-ferentiated South American source populations althoughfurther data to test this model are needed

To this end a thorough assessment of nuclear variationis warranted covering the same range as for the cpDNAwith particular attention to the northern contact zone Ananalysis of variation in the nuclear genome would alsoindicate whether populations had experienced a bottleneckin the past and thereby help to refute or verify the colon-ization hypothesis It would be useful to assess variationusing a phylogenetically informative nuclear marker suchas an intron sequence to permit preparation of a genealogyand an intraspecific phylogeographic hypothesis Clearlyit will be necessary to extend the study to cover the wholerange of the species including South America and theCaribbean The significance of the differentiation withinMesoamerican populations for both chloroplast and nuclearmarkers could then be assessed in a rangewide contextFinally it would be helpful to extend the phylogeographyof Mesoamerica to cover additional tree species and poten-tially other organisms This would shed light on any sharedpatterns of the colonization of Mesoamerica and contributeto the understanding of tropical forest response to chang-ing environment

Acknowledgements

The work reported in this paper was conducted as part of the EUproject lsquoAssessment of levels and dynamics of intraspecific geneticdiversity of tropical treesrsquo (contract ERBIC18CT970149 httpwwwnbuacukinco) which was coordinated by A Lowe Fund-ing for the work which contributed to the PhD thesis of S Caversis gratefully acknowledged from the EU and NERC supportedby the Centre for Ecology and Hydrology Edinburgh The authorswish to thank Marvin Hernandez Lionel Coto Sheila Ward andKevyn Wightman for assistance in obtaining samples and JuliaWilson (CEH) and Richard Ennos (University of Edinburgh) forconstructive contributions to the manuscript

References

Avise JC Arnold J Ball RM et al (1987) Intraspecific phylogeo-graphy the DNA bridge between population genetics and

systematics Annual Review of Ecology and Systematics 18 489ndash522

Bawa K Perry DR Beach JH (1985) Reproductive biology oftropical lowland rainforest trees I Sexual systems andincompatibility mechanisms American Journal of Botany 72331ndash345

Bermingham E Martin PA (1998) Comparative mtDNA phylo-geography of neotropical freshwater fishes testing sharedhistory to infer the evolutionary landscape of lower CentralAmerica Molecular Ecology 7 499ndash517

Bermingham E Moritz C (1998) Comparative phylogeographyconcepts and applications Molecular Ecology 7 367ndash369

Boshier DH (1984) The international provenance trial of Cordiaalliodora (R amp P) Oken in Costa Rica In Provenance and GeneticImprovement Strategies in Tropical Forest Trees Proceedings of a JointWork Conference Mutare Zimbabwe April 1984 (eds Barnes RDGibson GL) pp 168ndash187 Department of Forestry University ofOxford Oxford

Burnham RJ Graham A (1999) The history of neotropical vegetationnew developments and status Annals of the Missouri BotanicalGarden 86 546ndash589

Caron H Dumas S Marque G et al (2000) Spatial and temporaldistribution of chloroplast DNA polymorphism in a tropicaltree species Molecular Ecology 9 1089ndash1098

Chamberlain JR (1998) Isozyme variation in Calliandra calothyrsus(Leguminosae) its implications for species delimitation andconservation American Journal of Botany 85 37ndash47

Chaplin GE (1980) Progress with provenance exploration and seedcollection of Cedrela spp In Proceedings of the 11th CommonwealthForestry Conference pp 1ndash17 Commonwealth Forestry InstituteOxford

Chase MR Boshier DH Bawa KS (1995) Population genetics ofCordia alliodora (Boraginaceae) a neotropical tree 1 Geneticvariation in natural populations American Journal of Botany 82468ndash475

Coney PJ (1982) Plate tectonic constraints on the biogeographyof Middle America and the Caribbean region Annals of theMissouri Botanical Garden 69 432ndash443

Degen B Petit RJ Kremer A (2001) SGS mdash Spatial Genetic Softwarea computer program for analysis of spatial genetic and pheno-typic structures of individuals and populations Journal of Heredity92 447ndash449

Demesure B Sodzi N Petit RJ (1995) A set of universal primers foramplification of polymorphic non-coding regions of mitochon-drial and chloroplast DNA in plants Molecular Ecology 4 129ndash131

Dumolin-Lapegue S Demesure B Fineschi S Le Corre V Petit RJ(1997a) Phylogeographic structure of white oaks throughout theEuropean continent Genetics 146 1475ndash1487

Dumolin-Lapegue S Pemonge M-H Petit RJ (1997b) An enlargedset of consensus primers for the study of organelle DNA inplants Molecular Ecology 6 393ndash397

Dutech C Maggia L Joly HI (2000) Chloroplast diversity inVouacapoua americana (Caesalpinaceae) a neotropical foresttree Molecular Ecology 9 1427ndash1432

El Mousadik A Petit RJ (1996) Chloroplast phylogeography of theargan tree of Morocco Molecular Ecology 5 547ndash555

Ennos RA Sinclair WT Hu X-S Langdon A (1999) Using organellemarkers to elucidate the history ecology and evolution of plantpopulations In Molecular Systematics and Plant Evolution (edsHollingsworth PM Bateman RM Gornall RJ) pp 1ndash19 Tayloramp Francis Ltd London

P H Y L O G E O G R A P H Y O F C E D R E L A O D O R A T A 1459

copy 2003 Blackwell Publishing Ltd Molecular Ecology 12 1451ndash1460

Excoffier L Smouse PE (1994) Using allele frequencies andgeographic subdivision to reconstruct gene trees within aspecies molecular variance parsimony Genetics 136 343ndash359

Gentry AH (1982a) Phytogeographic patterns as evidence for aChoco refuge In Biological Diversification in the Tropics (edPrance GT) pp 112ndash136 Columbia University Press NewYork

Gentry AH (1982b) Neotropical floristic diversity phyto-geographical connections between Central and SouthAmerica Pleistocene climatic fluctuations or an accident ofthe Andean orogeny Annals of the Missouri Botanical Garden 69557ndash593

Gillies ACM Cornelius JP Newton AC et al (1997) Geneticvariation in Costa Rican populations of the tropical timberspecies Cedrela odorata L assessed using RAPDs MolecularEcology 6 1133ndash1145

Gillies ACM Navarro C Lowe AJ et al (1999) Genetic diversityin mesoamerican populations of mahogany (Swietenia macro-phylla) assessed using RAPDs Heredity 83 722ndash732

Graham A (1991) Studies in neotropical paleobotany X ThePliocene communities of Panama mdash composition numericalrepresentations and paleocommunity paleoenvironmentalreconstructions Annals of the Missouri Botanical Garden 78 465ndash475

Graham A (1999) Studies in Neotropical paleobotany XIII AnOligo-Miocene palynoflora from Simojovel (Chiapas Mexico)American Journal of Botany 86 17ndash31

Gregorius HR (1978) The concept of genetic diversity and itsformal relationship to heterozygosity and genetic distanceMathematical Bioscience 41 253ndash271

Hamilton MB (1999) Four primer pairs for the amplification ofchloroplast intergenic regions with intraspecific variationMolecular Ecology 8 521ndash522

Hewitt G (2000) The genetic legacy of the Ice Ages Nature 405907ndash913

Islebe GA Hooghiemstra H (1997) Vegetation and climate historyof montane Costa Rica since the last glacial Quaternary ScienceReviews 16 589ndash604

King RA Ferris C (1998) Chloroplast DNA phylogeography ofAlnus glutinosa (L) Gaertn Molecular Ecology 7 1151ndash1161

Lamb AFA (1968) Fast Growing Timbers of the Lowland Tropics No2 Cedrela odorata L Commonwealth Forestry Institute Univer-sity of Oxford Oxford

Lavin M Mathews S Hughes C (1991) Chloroplast DNAvariation in Gliricidia sepium (Leguminosae) intraspecificphylogeny and tokogeny American Journal of Botany 78 1576ndash1585

Leyden BW (1984) Guatemalan forest synthesis after Pleistocenearidity Proceedings of the National Academy of Sciences USA 814856ndash4859

Myers N Mittermeier RA Mittermeier CG da Fonseca GABKent J (2000) Biodiversity hotspots for conservation prioritiesNature 403 853ndash858

Navarro C Ward S Hernandez M (2002) The tree Cedrela odorata(Meliaceae) a morphologically subdivided species in CostaRica Revista Biologia Tropical 50 21ndash29

Newton AC Cornelius JP Mesen JF Leakey RRB (1995) Geneticvariation in apical dominance of Cedrela odorata seedlings inresponse to decapitation Silvae Genetica 44 146ndash150

Newton AC Allnutt TR Gillies ACM Lowe AJ Ennos RA

(1999a) Molecular phylogeography intraspecific variation andthe conservation of tree species Trends in Evolution and Ecology14 140ndash145

Newton AC Watt AD Lopez F et al (1999b) Genetic variation inhost susceptibility to attack by the mahogany shoot borerHypsipylla grandella (Zeller) Agricultural and Forest Entomology1 11ndash18

Pennington TD Styles BT Taylor DAH (1981) A Monographof the Neotropical Meliaceae New York Botanical GardensNew York

Petit RJ Kremer A Wagner DB (1993) Geographic structure ofchloroplast DNA polymorphisms in European oaks Theoreticaland Applied Genetics 87 122ndash128

Petit RJ Pineau E Demesure B et al (1997) Chloroplast DNA foot-prints of postglacial recolonisation by oaks Proceedings of theNational Academy of Sciences USA 94 9996ndash10001

Petit RJ Brewer S Bordacs S et al (2002a) Identification of refugiaand post-glacial colonisation routes of European white oaksbased on chloroplast DNA and fossil pollen evidence ForestEcology and Management 156 49ndash74

Petit RJ Csaikl UM Bordacs S et al (2002b) Chloroplast DNAvariation in European white oaks Phylogeography and patternsof diversity based on data from over 2600 populations ForestEcology and Management 156 5ndash26

Pons O Petit RJ (1995) Estimation variance and optimal samplingof gene diversity I Haploid locus Theoretical and AppliedGenetics 90 462ndash470

Pons O Petit RJ (1996) Measuring and testing genetic differentia-tion with ordered versus unordered alleles Genetics 144 1237ndash1245

Prance GT (1982a) A review of the phytogeographic evidencesfor Pleistocene climate changes in the Neotropics Annals of theMissouri Botanical Garden 69 594ndash624

Prance GT (1982b) Forest refuges evidence from woodyangiosperms In Biological Diversification in the Tropics (edPrance GT) pp 137ndash158 Columbia University Press New York

Raspe O Saumitou-Laprade P Cuguen J Jacquemart A-L (2000)Chloroplast DNA haplotype variation and population differen-tiation in Sorbus aucuparia L (Rosaceae Maloideae) MolecularEcology 9 1113ndash1122

Raven PH Axelrod DI (1974) Angiosperm biogeography and pastcontinental movements Annals of the Missouri Botanical Garden61 539ndash673

Rodan BD Newton AC Verissimo A (1992) Mahogany conserva-tion status and policy initiatives Environmental Conservation19 331ndash342

Savage JM (1982) The enigma of the Central American herpeto-fauna dispersals or vicariance Annals of the Missouri BotanicalGarden 69 464ndash547

Sewell MM Parks CR Chase MW (1996) Intraspecific chloroplastDNA variation and biogeography of North American Lirioden-dron L (Magnoliaceae) Evolution 50 1147ndash1154

Taberlet P Gielly L Pautou G Bouvet J (1991) Universal primersfor amplification of three non-coding regions of chloroplastDNA Plant Molecular Biology 17 1105ndash1109

Toledo VM (1982) Pleistocene changes of vegetation in tropicalMexico In Biological Diversification in the Tropics (ed PranceGT) pp 93ndash111 Columbia University Press New York

Tremblay NO Schoen DJ (1999) Molecular phylogeography ofDryas integrifolia glacial refugia and postglacial recolonisationMolecular Ecology 8 1187ndash1198

1460 S C A V E R S C N A V A R R O and A J L O W E

copy 2003 Blackwell Publishing Ltd Molecular Ecology 12 1451ndash1460

Valera FP (1997) Genetic Resources of Swietenia and Cedrela in theNeotropics Proposals for Coordinated Action Forest ResourcesDivision Forestry Department Food and Agriculture Organ-isation of the United Nations Rome

Williams M Dunkerley D de Deckker P Kershaw P Chappell J(1998) Quaternary Environments Arnold London

Wolfe KH Li W-H Sharp PM (1987) Rates of nucleotide substitu-tion vary greatly among plant mitochondrial chloroplast andnuclear DNAs Proceedings of the National Academy of SciencesUSA 84 9054ndash9058

Wright S (1969) Evolution and the Genetics of Populations Vol 2 TheTheory of Gene Frequencies University of Chicago Press Chicago

Stephen Cavers works at CEH minus Edinburgh on projects examininggenetic diversity and population structure in tropical tree speciesHe has completed a PhD at the University of Edinburgh of whichthis work forms part Carlos Navarro works at CATIE (Costa Rica)on the use of quantitative and molecular markers in conservationof tropical trees and forests He is currently studying at theUniversity of Helsinki Andrew Lowe runs the ecological geneticslaboratory at CEH mdash Edinburgh He is coordinator of several projectsaiming to apply genetic markers to the management of geneticresources for a range of tropical and temperate tree species