Biogeography of Apis cerana F. and A. nigrocincta Smith

[20] summarized the data then available onmorphometric variation in A. cerana. Herecognized four subspecies: A. c. cerana innorthern Asia; A. c. indicain southern Asia,A. c. japonicain Japan; and A. c. himalayain the Himalayan region.

I. INTRODUCTION

The Asian cavity-nesting honey bee Apisceranais widespread in temperate and trop-ical Asia. In his 1988 monograph, Ruttner

Original article

Biogeography of Apis ceranaF. and A. nigrocinctaSmith: insights from mtDNA studies

Deborah R. SMITHa*, Lynn VILLAFUERTEb, Gard OTISc,Michael R. PALMERa

a Department of Entomology, Haworth Hall, University of Kansas, Lawrence, KS 66045, USAb Institute of Biological Sciences, Genetics and Molecular Biology Division,

University of the Philippines, Los Banos, Philippinesc Department of Environmental Biology, University of Guelph, Guelph,

Ontario, N1G 2W1, Canada

(Invited paper)

Abstract –This study adds new data from Korea and the Philippines to earlier mitochondrial DNA(mtDNA)-based studies of the phylogeography of Asian cavity-nesting honeybees. A non-coding regionthat lies between the leucine tRNA gene and the cytochrome oxidase II gene of the mitochondrialgenome was sequenced in bees from 153 colonies of Apis cerana and A. nigrocincta, revealing 41 dif-ferent haplotypes. Five sequences could not be aligned with the others, two (from India and SriLanka) because the sequences were exceedingly A+T rich, and three (from Taiwan, the Philippines,and A. nigrocincta) because most of the non-coding sequence was absent. The remaining 36 sequenceswere aligned, and used in a phylogenetic analysis of A. ceranaand A. nigrocinctapopulations. Bothneighbor-joining and parsimony analyses were carried out, yielding similar results. We found five majorgroups of haplotypes: an Asian mainland group, a Sundaland group, a Palawan group, a Luzon-Mindanao group, and A. nigrocincta. The geographic distribution of these mtDNA haplotypes appearsto be strongly influenced by changes in sea-level during Pleistocene glaciations.

Apis cerana/ A. nigrocincta/ mtDNA / biogeography / phylogeny

Apidologie 31 (2000) 265–279 265© INRA/DIB-AGIB/EDP Sciences

* Correspondence and reprintsE-mail: [email protected]

D.R. Smith et al.266

However, our view of the biogeographyof A. ceranaand other Asian cavity-nest-ing species is undergoing changes as addi-tional samples and data become available.For example, Peng et al. [19] summarizedstudies showing that morphometric varia-tion exists among Chinese populations ofA. cerana. Damus [2] and Damus and Otis[3] carried out a morphometric analysis ofcavity-nesting bees from southeast Asia −a region from which few samples wereavailable to Ruttner [20]. They divided Rut-tner’s geographically widespread A. c. indicainto A. c. indica(samples from Sri Lanka),A. c. javana(samples from peninsularMalaysia, Borneo, Java, Bali, Lombok andFlores), a Timor population, and A. c. philip-pina from Luzon and Mindanao (subspe-cific names follow those proposed by Maa[15]). Smith and Hagen [23, 24] examinedgeographic variation of A. ceranacollectedfrom numerous locations, using as a sourceof information the sequence of a non-codingintergenic region in the mitochondrialgenome [1]. Phylogenetic analysis of thesedata indicated a strong geographic structurein the distribution of mitochondrial DNA(mtDNA) haplotypes. The major groups rec-ognized were mainland Asia (including sam-ples from India, Nepal, northern Thailand,Hong Kong, Korea and Japan), Sundaland(including samples from Samui island,peninsular Malaysia, Java, Bali, Lombok,Flores, Timor and Borneo), Sulawesi,Indonesia (samples recognized as A. nigro-cincta) and a Philippine group (based onsamples from Luzon and Mindanao).

In this study, we examine the biogeog-raphy of A. ceranaand A. nigrocincta,extending the work reported in Smith andHagen [23, 24] to include new samples fromKorea and the Philippines. We interpret ourresults in the light of Pleistocene geogra-phy and present-day climatic regions. In sodoing, we hope to add to the monumentalwork begun by Prof. Dr. Friedrich Ruttner,and contribute to a more complete under-standing of the intra- and inter-specific bio-geography and phylogeny of Apisspecies.

2. MATERIALS AND METHODS

2.1. Collections

Bees for this study were collected by theauthors, or donated by colleagues; Figure 1and Table I show approximate collectionsites. These samples consisted of workersfrozen in liquid nitrogen or preserved in~80% ethanol. Ten colony samples fromKorea were kindly provided by Dr. JaeChoe; these were from Inje’gun and Hong-chon’gun, Kangwon-do; Suwon City(2 samples), Kyonggi-do; Chechon City andYesangun, Chungchong-buk-do; MungyongCity and Youngchon’gun, Kyongsan-buk-do; Chongju City and Sunchang’gun,Cholla-buk-do; and Kochang’gun, andKyongsan-nam-do.

Older samples from the Philippines werecollected by Otis or Reyes. New samplesfrom the Philippines came from the collec-tion of the University of the Philippines,Los Banos Bee Program (UPLB). As such,collection times varied, with some samplesseveral years older than the rest. These sam-ples were examined earlier using morpho-metric techniques [26], and more recentlyusing DNA sequencing techniques [27].Philippine A. ceranaare from the follow-ing locations (see Fig. 2). Palawan island:1) Puerto Princessa, 2 colonies; 2) Brooke’sPoint, 3 colonies; 3) Quezon, 2 colonies;4) Roxas, 1 colony; 5) Taytay, 1 colony;6) El Nido, 1 colony from the UPLB; and7) Aborlan, 1 colony collected by Reyes.Luzon island, highlands: 8) BenguetProvince, 2 colonies from Baguio City,1 colony from Tublay, from the UPLB, and1 colony collected by G. Otis. Luzon island,lowlands: 9) Ilocos Norte Province, 4 colo-nies, 2 from Lubbot, Batac, 1 from SanMateo, 1 from San Pedro, Lumbao;10) Laguna Province, 3 colonies from theUPLB, 5 colonies collected by Otis;11) Zambales, 1 colony collected by Otis.Cebu island: 12) Argao, 1 colony. Panayisland: 13) Iloilo, 1 colony. Negros island:

Biogeography of Apis ceranaand A. nigrocincta

5’-GATCAATATCATTGATGACC-3’ [9].We sequenced the non-coding region usingthe internal primer 5’-GGCAGAATAAGT-GCATTG -3’ [1]. Autoradiograms were readby hand, and sequences were aligned byhand and with CLUSTAL W [14].

PAUP [25] was used for phylogeneticanalysis of 36 non-coding sequences. Fivesequences (IndiaY1, IndiaY2, SulawesiShort,TaiwanShort and PhilippineShort) wereomitted from phylogenetic analysis becausethey could not be aligned with the others.Substitutions among all four bases and inser-tion/deletions were weighted equally. Wecarried out a heuristic search for the mostparsimonious trees using the tree bisection-resection method, with 1 000 replicatesusing random addition of sequences. Wethen took the 50% majority consensus of allequally parsimonious trees. We also usedPAUP to construct a neighbor-joining tree[21] of these 36 sequences.

14) Negros Oriental, 2 colonies from BarrioBangbang and Bario Valencia; and 15)Negros Occidental, 1 colony from BarrioBanca, Kabangkalan. Leyte island: 16)Isabel, 1 colony. Mindanao island: 17) DavaoCity, 2 colonies; 18) Davao del Norte,1 colony; 19) Davao del Sur, 1 colony;20) North Cotabato, 2 colonies from Kida-pawan; and 21) Ozamis, 1 colony.

More detailed collection information onthe other samples is presented in Smith [22]and Smith and Hagen [23].

2.2. Sequencing studies

Details of DNA preparation and man-ual sequencing methods are provided else-where [18, 23]. We amplified a portion ofthe mitochondrial genome from the 3’endof COI to the 5’ end of COII using the primers5’-TCTATACCACGACGTTATTC-3’ and

267

Figure 1.A. cerana andA. nigrocincta collectingsites.

D.R. Smith et al.268

Table I. A. ceranaand A. nigrocinctacollections used in this study.

Location Haplotypes found (No. found)

Nepal Nepal1 (1)

India IndiaB1 (1), IndiaB2 (1), IndiaB3 (1), IndiaB4 (2),IndiaY1 (3)

Sri Lanka IndiaY2 (1)

Thailandnorthern Japan1 (2), Thai1 (3)Samui Island KoSamui1

Korea Nepal1 (1), Japan1 (7), Korea4 (1), Korea7 (1),Korea8 (1)

Japan Japan1 (14), Japan2 (1)

Hong Kong Japan1 (2)

Taiwan Taiwan1 (5)*

Malaysiapeninsula Malay1 (6), Malay2 (1), Malay3 (1)Borneo Borneo1 (1), Borneo2 (1), Borneo3 (1)

IndonesiaJava Java1 (7), Java2 (1)Bali Java1 (1), Bali1 (1), Bali2 (1), Bali3 (1)Lombok Lombok1 (5)Flores Java1 (4)Timor Java1 (4)Sulawesi Java1 (10), SulawesiY1 (2),

SangiheY1 (4), SulawesiShort (6)*Sangihe SangiheY1 (2), SulwesiShort (1)*

PhilippinesPalawan Palawan1 (4), Palawan2 (7)Luzon Luzon1 (13), Luzon2 (4), Cebu Cebu1 (1)Negros Negros1 (1), PhilippineShort (2)*Leyte Mindanao2 (1)Mindanao Mindanao1 (2), Mindanao2 (4), MindanaoP (1),

MindanaoL (1), Luzon2 (1)

Total number of specimens sequenced: 153.

A. nigrocintasamples found on the islands of Sulawesi and Sangihe, Indonesia, are indicated in boldface smallcapitals. Haplotypes indicated by an asterisk* are those in which most of the non-coding region is absent. Collectioninformation is given in the text for samples from the Philippines and Korea, and in Smith and Hagen [23] for otherlocations. Haplotype abbreviations are as in Figure 3.


We were able to align all but five of the 41non-coding sequences. Sequences IndiaY1and IndiaY2 from India and Sri Lanka (seeSmith and Hagen [23]) were found in beeslocally described as belonging to a ‘yellow’or ‘plains’ race. These two sequences areslightly shorter than most of the others, andthe base composition is almost 100% A+T.Although the 5’ and 3’ ends of these sequen-ces can be matched with the correspondingregions of the other sequences, the midregion cannot be aligned with confidence.This does not necessarily mean that the beescarrying the IndiaY1 or IndiaY2 haplotypes

3. RESULTS

Four haplotypes were found in new sam-ples from Korea (Japan1, Korea4, Korea7,Korea9) and 11 in the samples from thePhilippines (Palawan1, Palawan2, Cebu1,MindanaoP, Luzon1, Luzon2, MindanaoL,Mindanao11, Mindanao2, Negros1 andPhilippineShort). With the haplotypes pub-lished earlier [23], this totals 41 differentnon-coding sequences among 153 coloniessampled. Each sequence was given a shortname, indicating the locality in which it wasfirst observed.

269

Figure 2. A. cerana col-lecting sites in thePhilippines. Solid linesshow modern coastlines. Shading showsinferred late Pleistocenecoast lines, after Heaney[11]. Dots are approxi-mate locations of col-lection sites.

D.R

. Sm

ith et al.270

Figure 3.Non-coding sequences of A. ceranaand A. nigrocincta. Sequences of A. ceranafrom Korea and the Philippines, A. nigrocincta, and short hap-lotypes lacking most of the non-coding sequences are shown. See Smith and Hagen [23] for a complete listing and alignment of earlier sequences.Sequences are named after the first place they were found. Dashes indicate inferred insertion/deletion events.


Malay1-3, Borneo1-3, Java1-2, Bali1-3, andLombok1; a ‘Palawan’ group, includingPalawan1-2, MindanaoP (so called becauseit was found on Mindanao, but is similar tohaplotypes found on Palawan), and Cebu1;an A. nigrocincta group including Sulawe-siY1 and SangiheY1; and a Luzon-Min-danao group, including Luzon1-2, Min-danao1-2, MindanaoL (so called because itwas found on Mindanao, but is similar tosequences found on Luzon), and Negros1.

The heuristic search for the most parsi-monious midpoint-rooted tree of non-codingsequences produced 1 017 equally parsi-monious trees of length 80. The large num-ber of equally parsimonious trees is due tothe inclusion of sets of ‘nearly identical’sequences in the analysis (such as Bali1, 2,and 3, or Borneo1, 2, and 3; Smith andHagen [23]). The 50% majority rule con-sensus of these trees, shown in Figure 5,

are more distantly related to the otherA. cerana; we are simply unable to inter-pret the historical information that might becontained in these sequences. Three haplo-types − SulawesiShort, TaiwanShort andPhilippineShort − lack most of the non-cod-ing region. Only the 5’ and 3’ ends and afew bases between them are present. Fig-ure 3 shows the sequences found in Koreaand the Philippines, along with an A. nigro-cinctahaplotype, and the three haplotypesthat lack most of the non-coding region.

We excluded the sequences IndiaY1,IndiaY2, and the three short sequences fromthe phylogenetic analyses. The neighbor-join-ing gene tree for the remaining 36 sequen-ces is shown in Figure 4. This tree shows 5major branches: a ‘mainland’ group of hap-lotypes, including IndiaB1-4, Nepal1,Japan1-2, Korea4, 7 and 9 and Thai1;a ‘Sundaland’ group, including KoSamui1,

271

Figure 4. Neighbor-joiningtree based on the sequencesof 36 non-coding intergenicregions in the mtDNA ofA. ceranaand A. nigrocincta.Haplotype/sequence namesare as in Figure 3 and inSmith and Hagen [23].

D.R. Smith et al.

shows the same major groupings as theneighbor-joining tree. The neighbor-join-ing and majority-rule trees differ primarilyin the placement of A. nigrocincta(haplo-types SulawesiY1 and SangiheY1; Smithand Hagen [23]). The neighbor-joining treeunites A. nigrocinctahaplotypes with main-land Asia, Sundaland and Palawan A. ceranahaplotypes, while the 50% majority-ruleparsimony tree unites A. nigrocinctawiththe Luzon and Mindanao A. ceranahaplo-types.

Figure 6a shows the neighbor-joiningtree, but gives the geographic locations inwhich each haplotype was found. Figure 6bshows the main branches of the neighbor-joining tree and the biogeographic region(or species, in the case of A. nigrocincta) inwhich the sequences were found.

Short sequences, lacking most of the non-coding region, were found in Taiwan

(sequence TaiwanShort), Sulawesi andSangihe (sequence SulawesiShort) and inthe Philippine islands of Negros and Panay(sequence PhilippineShort). Superficially,it may seem that loss of most of the non-coding region is a synapomorphy unitingthese three sequences, but we feel it is morelikely that these are three independent losses.Cornuet et al. [1] sequenced this non-codingregion in A. mellifera, A. ceranafrom SriLanka, A. dorsataand A. florea. Theyshowed that the non-coding sequence in theSri Lankan A. cerana could be folded into aclover-leaf structure, the stem of which ismade by base-pairing between the conserved5’ and 3’ ends of the non-coding sequence.These stem sequences are conserved in allthe Apis species examined by Cornuet et al.[1] and in virtually all of the A. ceranaandA. nigrocinctawe examined. In the shortsequences, the 5’ and 3’ ends comprisingthe ‘stem’ are retained, but most of the

272

Figure 5. Fifty percentmajority rule consensus of1 017 equally parsimoniousmidpoint rooted trees of80 steps. Trees were gener-ated by a heuristic searchusing PAUP* [25] withthe following conditions:35 informative characters,tree bisection/resection,assignment of states to inter-nal nodes limited to statesobserved in the terminaltaxa; 1 000 replicates start-ing with random trees andrandom sequence addition.Numbers on branches indi-cate the frequency of thisgrouping in the 80 most par-simonious trees. Haplotypenames are as in Figure 3 andin Smith and Hagen [23].


4. DISCUSSION

This is part of an ongoing study of thebiogeography of Apis. Our conclusionsabout diversity and biogeography of A.ceranapopulations are limited by the

‘leaflets’ of the clover leaf have been lost(Fig. 7). The stem sequences can base pair toform a hairpin, with a stem and small loop.The exact points at which the ‘leaflets’ arelost, and the sequences left in the small loopsdiffer among the three short sequences.

273

Figure 6. a: Neighbor-joining tree (from Fig. 4)showing the locations inwhich each haplotype wasfound. b: Main branches ofneighbor-joining tree ofA. ceranahaplotypes andthe geographic region towhich each branch corre-sponds.

D.R. Smith et al.

samples available to us. Nonetheless, majorbiogeographic patterns within A. cerana arenow evident, and our results indicate fourgroups of mtDNA haplotypes within A.cerana: a mainland Asia group, a Sunda-land group, a Palawan group, and as Luzon-Mindanao group. A fifth group of haplo-types is found in our A. nigrocinctasamples.

The mainland Asia and Sundaland groupsare discussed in more detail in Smith andHagen [23], but several points are worthnoting here. First, India has an extremelydiverse collection of A. ceranahaplotypes,second only to the Philippines. These beesclearly merit further study. Second, themainland Asia group of mitochondrial hap-lotypes is extremely widespread. It occurs

over a larger geographic region than any ofthe other groups does, and there is little vari-ation among mainland haplotypes outsideof India. This suggests a large effective pop-ulation size and gene flow among theA. ceranapopulations of mainland Asia,perhaps coupled with rapid post-glacialrecolonization of the northern parts ofA. cerana’s range.

The Sundaland group includes mtDNAhaplotypes found in our A. ceranasamplesfrom peninsular Malaysia and the islandslying on the broad Sunda continental shelfof southeast Asia: the modern islands ofSumatra, Java, Bali, Borneo and manysmaller islands [10−12]. During Pleistoceneepisodes of glaciation, accumulation ofwater in glaciers and polar icecaps producedsea levels approximately 160 m lowerthan at present during the mid Pleistocene(160 000 years ago) and 120 m lower than atpresent in the late Pleistocene (16 000 to18 000 years ago [13]. During these times,the islands on the Sunda shelf would havebeen united directly with the Asian main-land by dry land, forming a region known asSundaland. Migration and gene flowbetween continental Asia and Sundalandwas followed by isolation of Sundaland pop-ulations as sea levels rose to their presentlevels.

The biogeographic break between theAsian mainland group and the Sundalandgroup lies in the Isthmus of Kra, in theMalay Peninsula. A recent study of mtDNArestriction site polymorphisms in A. ceranafrom Thailand [4] refines this observation.The authors report that populations ofA. cerana north of approximately 10°N lat-itude (in the Isthmus of Kra) show signifi-cantly different mtDNA haplotype fre-quencies from populations to the South. (Inearlier studies by Smith [22] and Smith andHagen [23], the samples described as beingfrom North and South Thailand were fromthe Chiang Mai and Bangkok regions, res-pectively; both sites are within Deowanishet al.’s North Thailand region.) In the work

274

Figure 7. The non-coding sequence of SriLankan A. cerana, and the three short sequences:A. ceranaTaiwanShort and PhilippineShort, andA. nigrocincta SulawesiShort, showing how theycan be folded into clover-leaf and hairpin struc-tures. The clover-leaf structure of Sri LankanA. ceranais redrawn from Cornuet et al. [1].


Palawan may have appeared as landabove the sea in the late Miocene or Pliocene.Today Palawan is separated from nearbyBorneo by a trench 145 m deep. During themid and late Pleistocene, the islands of thePalawan island chain would have beenunited into a larger island – Greater Palawan.Greater Palawan was united with the Asianmainland through Borneo in the mid Pleis-tocene, when sea levels were 160 m lowerthan present, but not in the late Pleistocene,when sea levels were only 120 m lower thanpresent. The other Philippine islands havehad no above-water connection to the Asianmainland. The distribution of the dwarfhoney bees reflects these connections:A. andreniformisis found on both Borneoand Palawan, but is apparently absent fromthe oceanic islands of the Philippines.

During the mid and late Pleistocene peri-ods of low sea levels many of the oceanicislands of the Philippines were joined bydry land, forming larger units, or ‘megaislands’: Greater Luzon, Greater Mindanao(which includes modern Mindanao, Samar,Leyte, and Bohol), and Greater Negros-Panay (which includes modern Negros,Panay, Cebu and Masbate). In the mid Pleis-tocene Greater Luzon and Greater Mindanaomay have been joined, though this is notcertain. Today the trench separating Luzonand Samar is 140 m deep, but it is also aregion of geological uplift. Two islandchains may form ‘stepping stones’ betweenBorneo and the oceanic islands of the Philip-pines: the Palawan chain between Borneoand Mindoro, and the Sulu Archipelagobetween Borneo and Mindanao.

As was the case with the Sundaland sam-ples, the distribution of the 11 mtDNA hap-lotypes found in our Philippine A. ceranasamples seems to be strongly influenced byPleistocene geological history. The Palawangroup of haplotypes includes Palawan1,Palawan2, Cebu1 and MindanaoP. Both ourneighbor-joining and parsimony trees showthe Palawan sequences to be more closelyrelated to Sundaland and mainland Asian

presented here, Samui Island is the onlyThai collection site that falls into Deowanishet al.’s South group. The mtDNA haplotypefrom Samui Island is distinct from that ofnorthern Thai samples and is placed withthe Sundaland group.

Why does the boundary between themainland and Sundaland mtDNA groupsfall in the Malay peninsula, so that mtDNAhaplotypes from southern Thailand andpeninsular Malaysia group with Sundalandrather than with the mainland Asia haplo-types? The Bilauktaung mountain range,which forms part of the boundary betweenMyanmar (Burma) and Thailand in theMalay Peninsula, may provide a barrier togene flow. In addition, this region of theIsthmus of Kra is known as the Kra eco-tone, in which there is a shift from ever-green rainforest (South of the imaginary linejoining the cities of Kangar and Pattani) tomore seasonal, semi-evergreen forest [28].In addition to marking the boundary betweenthe mainland and Sundaland groups ofA. cerana, the Kra ecotone also correspondsto the southern boundary of A. floreaon theAsian mainland [16].

Sulawesi was apparently never connectedby dry land to Sundaland and continentalAsia. A. nigrocinctahas been found onSulawesi, where it is sympatric with A. cer-ana, and on Sangihe [2, 3, 5, 8]. The biologyof Sulawesi cavity-nesting bees has beendiscussed elsewhere [5−7, 17].

The Philippines are a particularly com-plicated and interesting region for biogeog-raphers. The information presented here onPhilippine geology and biogeography islargely drawn from Heaney [10−12] andHeaney and Rickart [13]. Although someislands (notably Palawan) include a smallamount of continental crust, most of theislands were formed de novo by volcanicand tectonic activity. The degree to whichislands were connected to the mainland (viaSundaland) or to each other was influencedprimarily by the changes in sea level thatalso affected the islands of the Sunda shelf.

275

D.R. Smith et al.

sequences than to other sequences from thePhilippines. Sequences Palawan1 andPalawan2 have only been found on Palawan,showing the distinctiveness of that island’sA. cerana population. Two related sequences,Cebu1 and MindanaoP, have been found onthe islands of Cebu and Mindanao, respec-tively.

Six mtDNA haplotypes make up thegroup that we have called Luzon-Mindanao(see Figs. 4 and 5). These are Luzon1,Luzon2,MindanaoL, Mindanao1, Min-danao2, and Negros1. The geographic dis-tribution of these sequences shows theremay have been gene flow or migrationbetween Luzon and Mindanao, and betweenMindanao and Negros-Panay. We havefound only haplotypes Luzon1 and Luzon2in Luzon. Haplotypes Mindanao1 and Min-danao2 were the most common in our sam-ples from Greater Mindanao (i.e., Leyte andMindanao), but one example of Luzon2 andone example of MindanaoL were also foundthere. One example of the Luzon-Mindanaohaplotype ‘Negros1’ was found on the islandof Negros, part of Pleistocene GreaterNegros-Panay.

The last Philippine sequence was notincluded in phylogenetic analyses: this isPhilippineShort, a sequence lacking mostof the non-coding region. Two examples ofthis were found in the islands of GreaterNegros-Panay, one on Panay and one onNegros. Table I and Figure 6 show the dis-tribution of mtDNA haplotypes in the Philip-pine islands.

The Pleistocene ‘mega islands’ of thePhilippines have been shown to correspondto faunal regions, reflected in the distribu-tions of mammal species and genera [11,12, 13]; these regions are also applicable toA. cerana. Our survey of sequences alsoindicates that the relatively young Negros-Panay region may have received colonistsfrom Luzon or Mindanao (evidenced by thesequence Negros1, which resembles Luzonsequences) and Palawan, perhaps via Min-danao (evidenced by sequences Cebu1 and

MindanaoP, both of which resemble Palawansequences). The haplotype PhilippineShortmay be unique to the islands of GreaterNegros-Panay. The complexity of thePhilippines’ honey bee fauna merits furtherstudy − more intensive sampling from theregions discussed here, and samples fromother faunal regions such as the island ofMindoro.

In summary, our studies of A. cerana indi-cate four major groups of mtDNA lineages.These are a mainland mtDNA lineage, andthree additional mtDNA lineages occurringon lands connected to the mainland for suc-cessively shorter periods of time: Sunda-land (connected to the mainland during themid and late Pleistocene), Palawan (con-nected only during the mid Pleistocene),and the oceanic islands of the Philippines(never connected tot the mainland). TheIndonesian island of Sulawesi too, was neverconnected to the mainland, and here arefound two exceedingly similar cavity-nest-ing bee species, A. nigrocinctaand A. ceranawith Sundaland mtDNA.

ACKNOWLEDGMENTS

We heartily thank all the people who helpedus collect samples, or who provided samples forthis study: Jae Choe, Michael Crosland, MartinDamus, Fred Dyer, Akey Hung, Hermann Pech-hacker, Stefan Reyes, Masami Sasaki,Tadaharu Yoshida, and particularly A. Tilde andthe University of the Philippines Bee Program.We also thank Robert Hagen for critical com-ments on the manuscript and Sharon Hagen forthe illustrations.

Note added in proof

De la Rua et al. (2000) also analyzed variationin the non-coding region of mtDNA of Philip-pine A. cerana, using a slightly different surveymethod (De la Rua P., Simon E.U., Tilde A.C.,Moritz R.F.A., Fuchs S., MtDNA variation inApis ceranapopulations from the Philippines,Heredity 84 (2000) 124–130). Sihanuntavonget al. (1999) examined mtDNA restrictionsite variation in A. cerana of Thailand

276


nie maximum). Il en est résulté différentsdendogrammes : le dendogramme de lafigure 4 est issu de l’algorithme de neighbor-joining, celui de la figure 5 de l’analyse deparcimonie. Les deux méthodes conduisentau même résultat : 5 branches principalesréparties ainsi : 1) groupe d’haplotypes ducontinent asiatique ; 2) groupe de Sunda-land ; 3) groupe de Sulawan ; 4) un groupecommun à Luzon et Mindanao ; et 5) haplo-types de nigrocinta. La figure 6 donne larépartition géographique de ces haplotypes.La répartition géographique des haplotypessemble avoir été fortement influencée parles changement globaux du niveau de la merau cours du Pléistocène et par les surfaces deterres émergées qui en ont résulté dans lesud est asiatique. Au cours du Pléistocènemoyen et supérieur, les îles du plateau conti-nental Sunda ont été unies les unes auxautres (ou n’étaient séparées que par deschenaux très étroits) et unies au continentasiatique. Palawan a été relié à Bornéo etau continent asiatique au cours du Pléisto-cène moyen mais pas au Pléistocène supé-rieur. Sulawesi et les îles océaniques dénom-mées Philippines n’ont jamais été reliées aucontinent asiatique.

Apis cerana / A. nigrocincta/ ADNmt /biogéographie / phylogenèse

Zusammenfassung – Biogeographie vonApis ceranaF. und Apis nigrocinctaSmith: Ergebnisse von mtDNA Untersu-chungen. Wir untersuchten die Variationder DNA Sequenz in einer nicht codierendenRegion des mitochondrialen Genoms von153 Völkern von Apis ceranaund A. nigro-cincta. Die Proben von A. cerana stammten ausIndien, Sri Lanka, Nepal, Thailand, China(Hong Kong), Taiwan, Korea, Japan, Malay-sia (Halbinsel und Borneo), den indonesi-schen Inseln Java, Bali, Lombok, WestTimor, Flores und Sulawesi (Abb. 1 undTab. I) und den philippinischen Inseln Pala-wan, Luzon, Leyte, Mindanao, Cebu, Panay

(Sihanuntavong D., Sittipraneed S., KlinbungaS., Mitochondrial DNA diversity and populationstructure of the honeybee (Apis cerana) in Thai-land, J. Apic. Res. 38 (1999) 211–219). Theresults of both studies agree with the data reportedhere.

Résumé – Biogéographie d’Apis ceranaF.et d’Apis nigrocinctaSmith : résultats desétudes d’ADNmt. Nous avons étudié lavariation de la séquence d’ADN dans unerégion non codante des génomes mito-chondriaux de 153 colonies d’A. ceranaet d’A. nigrocincta. Les échantillonsd’A. ceranaprovenaient d’Inde, du SriLanka, du Népal, de Thaïlande, de Chine(Hong Kong), de Taïwan, de Corée, duJapon, de la péninsule de Malaisie et de Bor-néo, des îles indonésiennes de Java, Bali,Lombok, Timor occidental, Flores et Sula-wesi (Fig. 1, Tab. I) et des îles Philippines dePalawan, Luzon, Leyte, Mindanao, Cebu,Panay et Negros (Fig. 2, Tab. I). Les échan-tillons de nigrocinctavenaient des îles indo-nésiennes de Sulawesi et de Sangihe.La figure 3 montre les séquences d’ADNmttrouvées dans les nouveaux échantillons deCorée (Japon1, Corée4, Corée7 et Corée9)et dans 11 échantillons des Philippines (Pala-wan1, Palawan2, Cebu1, MindanaoP,Luzon1, Luzon2, MindanaoL, Mindanao11,Mindanao2, Negros1 et PhilippineShort).Avec les haplotypes publiés précédemment[23] on arrive à un total de 41 séquencesdifférentes non codantes parmi les 153 colo-nies échantillonnées. Deux des 6 haplotypesobservés parmi les colonies d’Inde et du SriLanka n’ont pu être alignées avec les autresséquences d’A. cerana. La plus grande par-tie de la région non codante était absente de3 haplotypes (TaïwanShort, SulawesiShortet PhilippineShort) (Figs. 3 et 7). Laséquence non codante a probablement étéperdue indépendamment à trois reprises. L’analyse phylogénétique des 36 séquencesnon codantes alignées (les séquences courteset les deux séquences non alignées mises àpart) a été faite à l’aide de deux méthodesstatistiques (neighbor-joining et parcimo-

277

D.R. Smith et al.

und Negros (Abb. 2 und Tab. I). Die Pro-ben von A. nigrocincta stammten von denindonesischen Inseln Sulawesi und Sangihe.In Abbildung 3 sind die mtDNA Sequen-zen dargestellt, die in den neuen Proben vonKorea (Japan1, Korea4, Korea7, Korea9)und 11 in den Proben aus den Philippinen(Palawan1, Palawan2, Ceb1, MindanaoP,Luzon1, Luzon2, MindanaoL, Mindanao11,Mindanao2, Negros1 und Philippinenshort(kurz) gefunden wurden. Zusammen mit denHaplotypen, die schon früher publiziert wur-den [23], ergeben sich von den 153 beprob-ten Völkern 41 verschiedene nicht kodie-rende mitochondriale Sequenzen. Zwei von6 Haplotypen, die in Völkern aus Indien undSri Lanka gefunden wurden, konnten denanderen A. ceranaSequenzen nicht zuge-ordnet werden. Bei 3 Haplotypen (alle Pro-ben von Taiwan, einige Apis ceranaPro-ben von Negros und Panay, Philippinen undeinige A. nigrocincta)fehlte das gröβteStück der nicht kodierenden Region (Abb. 3und 7). Dieser Verlust der nicht kodierendenSequenzen ist wahrscheinlich dreimal unab-hängig voneinander entstanden.Phylogenetische Analysen von 36 zusam-menhängenden nicht kodierenden Sequen-zen (mit Ausnahme der kurzen Sequenzenund zwei anderen, die nicht zugeordnet wer-den konnten) wurden mit zwei verschiede-nen statistischen Methoden durchgeführt(neighbor-joining und maximum parsi-mony). Es ergaben sich verschiedene Den-drogramme: das Dendrogramm der Abbil-dung 4 entstand durch neighbor-joiningalgorithm, das in Abbildung 5 entstanddurch die 50% majority rule consensus ofall most parsimonious trees. Beide Verfah-ren führen zu 5 Hauptästen: 1) Gruppe vonHaplotypen vom asiatischen Festland;2) Sundaland Gruppe; 3) Palawan Gruppe;4) eine gemeinsame Luzon und MindanaoGruppe; und 5) A. nigrocincta Haplotypen.Die geographische Verteilung dieser Haplo-typen ist in Abbildung 6 zu sehen.Die geographische Verteilung scheint starkvon den weltweiten Änderungen des Was-serspiegels während des Pleistozäns und

den dadurch entstandenen Landflächenbeeinflusst worden zu sein. Während desmittleren und späten Pleistozäns waren dieInseln auf dem Sundariff miteinander (odernur durch sehr enge Kanäle getrennt) undmit dem Festland verbunden. Palawan hingmit Borneo zusammen, und während desmittleren Pleistozäns auch mit dem Fest-land, war aber im späten Pleistozän wiedergetrennt. Sulawesi und die sogenanntenOzean Inseln der Philippinen standen mitdem Festland nie in Verbindung.

Apis cerana/ A. nigrocincta / Mitochon-driale DNA / Biogeographie / Phylogenie

REFERENCES

[1] Cornuet J.-M., Garnery L., Solignac M., Puta-tive origin and function of the intergenic regionbetween COI and COII of Apis melliferaL. mito-chondrial DNA, Genetics 128 (1991) 393–403.

[2] Damus M.S., Morphometric and genetic analysisof cavity-nesting honey bees (Apis cerana F.) ofsoutheast Asia, Master’s thesis, Univ. Guelph,Canada, 1995.

[3] Damus M.S., Otis G.W., A morphometric anal-ysis of Apis ceranaF. and Apis nigrocinctaSmithpopulations from southeast Asia, Apidologie 28(1997) 309–323.

[4] Deowanish S., Matsuka M., Nakamura J.,A study of diversity of Apis ceranain Thailandusing PCR-RFLP of mtDNA, in: Schwartz P.,Hogendoorn K. (Eds.), Social Insects at theTurn of the Millennium, Proc. XIII Int. IUSSICongr., Flinders Univ. Press, Adelaide, Aus-tralia, p. 138.

[5] Hadisoesilo S., A comparative study of twospecies of cavity-nesting honey bees of Sulawesi,Indonesia, PhD thesis, Univ. Guelph, Canada,1997.

[6] Hadisoesilo S., Otis G.W., Drone flight timesconfirm the species status of Apis nigrocinctaSmith, 1861 to be a species distinct from ApisceranaF., 1793, in Sulawesi, Indonesia, Api-dologie 27 (1996) 361–369.

[7] Hadisoesilo S., Otis G.W., Differences in dronecappings of Apis ceranaand Apis nigrocincta,J. Apic. Res. 37 (1998) 11–15.

[8] Hadisoesilo S., Otis G.W., Meixner M., Mor-phometric analysis reveals two distinct cavity-nesting honey bees (Hymenoptera: Apidae) inSulawesi, Indonesia, J. Kans. Entomol. Soc. 68(1995) 339–407.

278


[19] Peng Y.S., Nasr M.E., Locke S.J., Geographi-cal races of Apis ceranaFabricius in China andtheir distribution. Review of recent Chinese pub-lications and a preliminary statistical analysis,Apidologie 20 (1989) 9–20.

[20] Ruttner F., Biogeography and taxonomy of hon-eybees, Springer-Verlag, Berlin, 1988.

[21] Saitou N., Nei M., The neighbor-joining method:a new method for reconstructing phylogenetictrees, Mol. Biol. Evol. 4 (1987) 406–425.

[22] Smith D.R., Mitochondrial DNA and honey beebiogeography, in: Smith D.R. (Ed.), Diversityin the Genus Apis, Westview Press, Boulder,CO, 1991, pp. 131–176.

[23] Smith D.R., Hagen R.H., The biogeography ofApis ceranaas revealed by mitochondrial DNAsequence data, J. Kans. Entomol. Soc. (suppl.) 69(1996) 249–310.

[24] Smith D.R., Hagen R.H., Phylogeny and bio-geography of Apis ceranasubspecies: testingalternative hypotheses, in: Hoopingarner R.,Connor L. (Eds.), Apiculture for the 21stCentury, Wicwas Press, Cheshire, CT, 1999,pp. 60–68.

[25] Swofford D.L., PAUP*4 Phylogenetic Analy-sis Using Parsimony (*and other methods), Ver-sion 4, Sinauer Associates, Sunderland, MA,1998.

[26] Tilde A., Morphometric diversity of Apis cer-ana Fabr. in the Philippines, in: Schwartz M.P.,Hogendoorn K. (Eds.), Social Insects at theTurn of the Millennium, Proc. XIIIth Int. Congr.IUSSI, Flinders Univ. Press, Adelaide, Australia,1998, pp. 60–68.

[27] Villafuerte L.S., Mitochondrial DNA (mtDNApolymorphism in the Asian honeybee (Apis cer-anaF.) in the Philippines, Ph.D. thesis, Univ.Philippines, Los Banos, 1999.

[28] Whitmore T.C., Tropical Rain Forests of the FarEast, 2nd ed., Oxford Sci. Publ., Clarendon Press,Oxford, 1984.

[9] Hall H.G., Smith D.R., Distinguishing Africanand European honeybee matrilines using ampli-fied mitochondrial DNA, Proc. Natl. Acad. Sci.U.S.A. 88 (1991) 4548–4552.

[10] Heaney L.R., Zoogeographic evidence for mid-dle and late Pleistocene land bridges to the Philip-pine Islands, Mod. Quat. Res. SE Asia 9 (1985)127–143.

[11] Heaney L.R., Biogeography of mammals in SEAsia: estimates of rates of colonization, extinc-tion and speciation, Biol. J. Linn. Soc. 28 (1986)127–165.

[12] Heaney L.R., A synopsis of climatic and vege-tational change in southeast Asia, ClimaticChange 19 (1991) 53–61.

[13] Heaney L.R.,. Rickart E.A., Correlations ofclades and clines: geographic, elevational andphylogenetic distribution patterns among Philip-pine mammals, in: Peters G., Hutterer R. (Eds.),Vertebrates In The Tropics, Museum Alexan-der Koenig, Bonn, 1990, pp. 321–332.

[14] Higgins D.G., Sharp P.M., CLUSTAL: a pack-age for performing multiple sequence alignmenton a microcomputer, Gene 73 (1988) 237–244.

[15] Maa T.-C., An inquiry into the systematics ofthe tribus Apidini or honeybees (Hym.), Treubia21 (1953) 525–640.

[16] Otis G., Distributions of requently recognizedspecies of honey bees (Hymenoptera: Apidae;Apis) in Asia, J. Kans. Entomol. Soc. 69 (1996)331–333.

[17] Otis G.S., Hadisoesilo S., Insights into honeybee biology from Apis nigrocinctaof Indone-sia, in: Hoopingarner R., Connor L.J. (Eds.),Apiculture for the 21st Century, Wicwas Press,Cheshire, CT, 1999, pp. 69–79.

[18] Palmer M.R., Smith D.R., Kaftanoglu O., Turk-ish honey bees, genetic variation and evidence fora fourth lineage of Apis melliferamtDNA,J. Hered. 91( 2000) 42–46.

279

Documents

Biogeography of Apis cerana F. and A. nigrocincta Smith