Content uploaded by Deborah Roan Smith
Author content
All content in this area was uploaded by Deborah Roan Smith on Mar 13, 2014
Content may be subject to copyright.
multivariate probability terms (Hepburn and
Radloff, 1998; Ruttner, 1988). Contempo-
rary classification of honeybees stems from
multivariate methods of analysis originally
advanced by DuPraw (1964, 1965) and
substantially developed by Ruttner (1988),
1. INTRODUCTION
The history of honeybee classification
reflects a slow movement away from the
fixed abstractions of the linnaean system to
the analysis of population dynamics in
Review article
Infraspecific categories of Apis cerana:
morphometric, allozymal and mtDNA diversity
H. Randall HEPBURNa*, Deborah R. SMITHb,
Sarah E. RADLOFFc, Gard W. OTISd
aDepartment of Zoology and Entomology, Rhodes University, Grahamstown 6140, South Africa
bDepartment of Entomology, University of Kansas, Lawrence, Kansas 66045, USA
cDepartment of Statistics, Rhodes University, Grahamstown 6140, South Africa
dDepartment of Environmental Biology, University of Guelph, Guelph, Ontario,
N1G 2W1, Canada
(Received 4 July 2000; accepted 29 September 2000)
Abstract – An analysis of the infraspecific categories of Apis cerana was prepared from the relevant
literature on taxonomy, morphometrics, allozyme polymorphism and mtDNA diversity. About 31 puta-
tive biometric groups have been proposed and assigned to about eight equivocal “subspecies” and var-
ious ecotypes. However nearly half of the area of distribution of A. cerana remains unexamined.
Allozyme polymorphism is greatest in southeast and lowest in northern and western Asia. About
four major mtDNA groups are discernable. There is a very low overall geographic congruity amongst
the morphoclusters, allozyme polymorphs and mtDNA clusters. The greatest problems in resolving
infraspecific categories in A. cerana are inadequate sampling, incompatible differences in sample sizes,
character suites, sampling distance, confidence limits and range of geographical scales employed
in different studies.
Apis cerana / taxonomy / biogeography / Asia / honeybees
Apidologie 32 (2001) 3–23 3
© INRA/DIB-AGIB/EDP Sciences, 2001
* Correspondence and reprints
E-mail: r.hepburn@ru.ac.za
H.R. Hepburn et al.
4
Ruttner et al. (1978) and Daly (1991, 1992).
The last decade has been particularly fruit-
ful in this regard. Ruttner (1988) completed
the daunting task of providing the first mul-
tivariate analytical attempt at a comprehen-
sive macroscale synthesis of honeybee clas-
sification for the genus Apis. This was a
major impetus for subsequent mesoscale
studies of honeybee morphometrics in
A. mellifera L. (cf. Hepburn and Radloff,
1998) as well as for the analysis of allozymic
and DNA diversity of honeybee populations
(Arias and Sheppard, 1996; Cornuet and
Garnery, 1991; Smith, 1991; Smith et al.,
1991).
In parallel with studies of A. mellifera,
the Ruttner monograph (1988) also opened
a new chapter in the study of the classifica-
tion and biogeography of the honeybees of
Asia. This is particularly evident in a recent
spate of Asian regional studies in several
journals and monographs (Verma, 1990,
1992). The purpose of this communication is
to report the results of a survey on the pub-
lished literature to date on the infraspecific
classification of A. cerana Fabr. throughout its
entire natural range of some 30 million km2.
The sympatric occurrence of other Apis
species with A. cerana in southeastern Asia
raises a very undesirable spectre: some pre-
vious “A. cerana” literature could well be
contaminated through the inadvertent inclu-
sion of data derived from species other than
A. cerana. The likelihood of detecting such
errors seems remote at best. We have
included the results of morphometric stud-
ies as well as those on allozymic and DNA
diversity. In the end we present a current
portrait of putative infraspecific categories
of A. cerana. There were no formal “Mate-
rials and Methods” for this study, solely
analyses of the published literature cited in
the references. It should be noted that the
approach taken was one of trying to resolve
distinct groups of A. cerana independently
of “correcting” their complex taxonomic
history in terms of the International Code
of Zoological Nomenclature (Engel, 1999).
2. RESULTS AND DISCUSSION
The natural distribution of A. cerana
based on published citations with reference
to specific localities is shown in Figure 1.
Literature consulted was predominantly that
recoverable from Apicultural Abstracts
(1950–1999). Positions of localities (closed
circles) are only approximate because of
map scale; areas in which A. cerana has
been specifically sought but found absent
are indicated with stars (Fig. 1). There are
several regions where A. cerana undoubt-
edly occurs but for which there is extremely
sparse or no data at all (Afghanistan, much
of India, Laos, Cambodia, Myanmar and
Sumatra) and other areas where it has
recently been introduced (Papua New
Guinea) but these are not considered.
2.1. Morphometrics
2.1.1. Western Asia
This region extends from the western bor-
ders of Afghanistan to the north and Pak-
istan to the south at about longitude 60°,
thence eastwards below the Himalayan
mountain range and across the Indian sub-
continent to Myanmar at about longitude
94° (Fig. 2). The extent and quality of infor-
mation on the honeybee populations of this
area (about 4 million km2) is extremely vari-
able and ranges from anecdotal descriptions
to full multivariate statistical analyses of
morphometric characters.
The only information on the classifica-
tion of the honeybees of Afghanistan and
Pakistan (Fig. 2, area 1) are those of Maa
(1953) and Ruttner (1988, 1992) who con-
cluded, respectively, that these bees could
not be morphologically or morphometri-
cally discriminated from those of neigh-
bouring China (classically regarded as
A. cerana cerana (Ruttner et al., 1989).
However, extremely few samples from this
area were available to Ruttner. Further pos-
sible discrimination of these honeybee
Infraspecific categories of Apis cerana
well as population structure for the honey-
bees of this region is blurred because of fun-
damental incompatibilities in the method-
ologies and forms of analysis employed in
studies published to date (Fig. 2, areas 2–3,
7–11, 14–16). This becomes particularly
evident from comparisons of three impor-
tant publications based on the Indian sub-
continent. Firstly, Kshirsagar (1983) pro-
posed some seven ecotypes for the bees
of India vis-a-vis the more usually accepted
“hills” and “plains” varieties or ecotypes
(Kapil, 1956; Narayanan et al., 1960,
1961a, b; Ruttner, 1988). Secondly, there is
a series of papers by Verma and colleagues
(Mattu and Verma, 1983a, b; 1984a, b;
Sihanuntavong et al., 1999; Singh et al.,
populations is suggested by a report of two
different “kinds” of bees from Afghanistan
even though the identifications made at the
time subsequently proved incorrect (Schnei-
der and Djalal, 1970). Thus, virtually noth-
ing is known about population structure of
the honeybees of Afghanistan and Pakistan.
Of greater interest, nothing is yet known of
the details of honeybees in eastern Iran
which is geographically the closest point of
contact between A. cerana and A. mellifera
(Ruttner, 1988; Ruttner et al., 2000).
The honeybees of the sub-Himalayan and
Indian regions have been intensively inves-
tigated in recent years. However, the current
picture of biometric groups and ecotypes as
5
Figure 1. Known distribution of A. cerana = (circles); reported absence = (stars).
H.R. Hepburn et al.
6
Figure 2. Distributional areas of putatively distinct subspecies, biometric groups and/or ecotypes of
A. cerana. A. cerana cerana: 1. Eastern Afghanistan and northern Pakistan (by inference only);
2. Kashmir; 3. Himachal Pradesh; 4. China (with biotypes/ecotypes a = Yunnan, b = Guangdong-
Guangxi, c = Hunan, d = northern, e = Changbei Shan, f = unspecified, g = Taiwan); 5. Korea;
6. Ussuria. A. cerana himalayana: 7. Nepal Terai plains; 8. Nepal midlands; 9. Himalayas;
10. Brahmaputra; 11. Manipur, Mizoram and Nagaland. A. cerana skorikovi: (possibly = A. cer-
ana cerana) 12. Tibet. A. cerana abaensis: 13. Central China. A. cerana indica: 14. Uttar Pradesh;
15. Orissa; 16. Southern India; 17. Sri Lanka (with montane, lowland and Anuradhadpura ecotypes);
18. Yunnan and possibly northern Myanmar; 19. Northern Thailand; 20. Southern Thailand and
continental Malaysia; 21. Phuket Island; 22. Samui Island; 23. Sumatra (northern half by inference
only), Java, Borneo, Lombok, Bali, Flores and most of Sulawesi; 24. Southern Sulawesi; 25. Timor;
26. Sabah. A. cerana hainanensis: 27. Hainan Island (with coastal and montane ecotypes).
A. cerana philippina: 28. Visayas and Mindanao; 29. Luzon (with highland and lowland ecotypes);
30. Palawan (distinguishable from other Philippine morphoclusters). A. cerana japonica: 31. Japan
(with two ecotypes). Undesignated areas on the map remain unknown. Map constructed from:
Akahira and Sakagami, 1959a, b; Avetisyan, 1960; Damus and Otis, 1997; Diniz-Filho et al., 1993;
Engel, 1999; Fernando, 1979; Fuchs et al., 1996; Hadisoesilo et al., 1995; Kapil, 1956; Kshirsagar,
1973, 1983; Kwon and Huh, 1992; Lawrjochin, 1960; Limbipichai, 1990; Maa, 1953; Mattu and
Verma, 1983a, b, 1984a, b; Muzaffar and Ahmad, 1989; Narayanan et al., 1960, 1961a, b; Ono,
1992; Otis, 1991; Otis and Hadisoesilo, 1990; Peng et al., 1989; Pesenko et al., 1989; Rinderer et al.,
1989; Ruttner, 1988, 1992; Ruttner et al., 1978, 1989; Sakai, 1956, 1958; Sasaki, 1994; Schneider and
Djalal, 1970; Schneider and Kloft, 1971; Singh et al., 1990; Sylvester et al., 1998; Tilde et al., 2000;
Tokuda, 1924; Verma, 1990, 1992; Verma et al., 1989, 1994; Yang, 1989; Zhen-Ming et al., 1992.
Infraspecific categories of Apis cerana
as the bees of southern Pakistan would group
together as populations of A. cerana indica
in the Ruttner system.
To complete the tally for western Asia,
the honeybees of Afghanistan and northern
Pakistan would presumably show close
affinities to A. cerana cerana. The three
ecotypes proposed for Sri Lanka (Fig. 2,
area 17) by Fernando (1979) apparently
coincide with mainland A. cerana indica
(Damus and Otis, 1997; Fuchs et al., 1996).
However, it must be noted that if the bees of
Sri Lanka are indeed to be named A. cer-
ana indica these “indica” are not the same
biometric “indica” found further east in
Malaysia (Damus and Otis, 1997). The cur-
rent picture of honeybee subspecies, bio-
metric groups, morphoclusters, ecotypes
and/or biotypes in western Asia (Fig. 2) can
only be regarded as highly suggestive and
tentative because they emanate from sepa-
rate studies which are not cross-compati-
ble.
2.1.2. Northeast Asia
Northeast Asia as here defined includes
China, the Manchurian plains of the former
USSR, Korea and Japan (Fig. 2). The hon-
eybees of the vast territory of China have
been systematically investigated in a lengthy
series of publications, principally by Yang
and colleagues (Peng et al., 1989; Yang,
1989) who reached a number of conclusions
based on analyses of honeybees from more
than 1000 localities. They proposed a series
of five major biometric groups or morpho-
clusters as well as several ecotypes (cf. Peng
et al., 1989). The same groups or races have
been subsequently supported (Zhen-Ming
et al., 1992). There are two aspects to this
work. Firstly, there are the original results of
the Yang group and secondly, some new
analyses and comments on parts of the same
database by Peng et al. (1989).
Figure 2 (areas 4, 12, 13, 18, 27) illus-
trates the distributions of the five honeybee
races that emanated from Yang’s group as
well as the several ecotypes within them.
1990; Verma et al., 1989, 1994) that began
with univariate methods but soon progressed
to multivariate analyses. Several important
points arise from the latter studies.
Singh et al. (1990) used a full suite of
multivariate techniques and identified three
biometric groups in the eastern Himalayan
region: 1. Manipuri bees from Nagaland,
Manipur and Mizoram (Fig. 2, area 11);
2. Brahmaputra bees from that valley and
also from southern Assam and Megahalaya
(Fig. 2, area 10); and 3. Himalayan bees
from Sikkim, West Bengal, northern and
western Assam and Arunachal Pradesh
(Fig. 2, area 9). In a complementary study
Verma et al. (1994) analysed bees from
Nepal and the western Himalayas and could
discriminate four biometric morphoclusters:
1. Terai plains bees in Nepal (Fig. 2, area 7);
2. Midland bees in the Nepali midlands
(Fig. 2, area 8); 3. Himachali bees from
Himachal Pradesh (Fig. 2, area 3); and
4. Kashmiri bees from Kashmir in northern
India (Fig. 2, area 2). How the honeybees
of Kashmir may relate to those of the north-
western frontier of Pakistan (Muzaffar and
Ahmad, 1989) cannot yet be determined.
The mesoscale analyses of Singh et al.
(1990) and Verma et al. (1994) were thor-
ough multivariate analyses but performed
on entirely unrelated, non-contiguous
databases. Therefore, these rigorous
mesoscale results are localised and cannot be
extrapolated or statistically amalgamated to
assess the whole region. In the circum-
stances this leaves a single macroscale anal-
ysis for the region in the study of Ruttner
(1988) whose work provides conclusions
on far less data and which is presented in a
way that precludes any further numerical
analysis. In any event, in the Ruttner per-
spective, the Kashmiri and Himachali bees
(Fig. 2, areas 2–3) would be classified as
A. cerana cerana while those extending
across Nepal to the border of Myanmar as
populations of A. cerana himalayana (Fig. 2,
areas 14, 7–11). The majority of the eco-
types proposed by Kshirsagar (1983) as well
7
H.R. Hepburn et al.
Peng et al. (1989) stated that the original
Chinese studies did not include sufficient
raw data nor details of descriptive statistics
with which to re-evaluate the findings.
Nonetheless, Peng and colleagues used mul-
tivariate methods to re-analyse some of the
original Yang data but using a small char-
acter suite of only three morphometric fea-
tures. So, there is an intrinsic difficulty both
in the interpretation of the original Yang
data as well as in the limited database pro-
cessed by Peng et al. (1989).
Although Peng et al. (1989) were unable
to support the biotypes of A. cerana haina-
nensis, they did however demonstrate a sig-
nificant discrimination function (but of low
probability) for the five biotypes of A. cer-
ana cerana proposed by Yang (Fig. 2, area
4a–e). Peng et al. (1989) concluded that
methodological differences between the
Yang group and others preclude compar-
isons of these putative groups (Fig. 2) with
those emanating from other honeybee stud-
ies in eastern Asia. There are only two pos-
sible secure links for this data. There appears
to be a safe link between the A. cerana
indica of southern Yunnan with the honey-
bees in the subtending Indochina peninsula
(this same point emerged from the
macroscale studies of Ruttner (1988, 1992).
Finally, while Peng et al. (1989) were able
to confirm the separateness of A. cerana
cerana and A. cerana skorikovi but not the
other three Yang races, they were inclined to
accept these other races on the basis of
behavioural and other biological character-
istics. A very limited amount of data from
the honeybees of China were available to
Ruttner (1988) and he was only able to state
that the bees of northern China was A. cerana
cerana and those of the southwest a differ-
ent, unspecified subspecies.
Although A. cerana is apparently non-
native to the great expanses of the former
Soviet Union, even in areas very near north-
ern Afghanistan such as Tajikistan and Khir-
gizia, it resurfaces in the far eastern region
of the Ussuri (or Primorsky) district (Fig. 2,
area 6) just eastwards of Manchurian China
(Avetisyan, 1960; Lawrjochin, 1960;
Pesenko et al., 1989). Although Lawrjochin
(1960) suggested that the Ussurian bees
were close to A. cerana japonica, the Rus-
sian literature does not appear to provide
any morphometric data on the bees of this
region. Ruttner (1988, 1992) apparently did
not have access to Ussurian honeybee sam-
ples and did not comment either way.
Although never published, A. cerana appar-
ently also occurs in eastern Mongolia
(Choon Thin Yat, personal communication).
The honeybees of peninsular Korea (Fig. 2,
area 5) have been analysed in a series of
papers by Kwon and colleagues (cf. refer-
ences in Kwon and Huh, 1992). Basically,
they studied samples from fifteen localities
in southern Korea and placed them all within
the same biometric group. In the absence of
re-analysable data, these results cannot be
compared with any other Asian work.
Ruttner (1988) seemed to regard these bees
as morphometrically intermediate between
A. cerana cerana of the mainland and
A. cerana japonica in Japan.
The honeybees of the islands of Japan
have been extensively analysed over the last
century. It is general consensus that these
bees are morphometrically completely iso-
lated from others of the A. cerana complex
(Damus and Otis, 1997; Ruttner, 1988;
Sasaki, 1994) as well as in terms of mtDNA
haplotypes (Deowanish et al., 1996). This
isolation provides a convenient basis for
studies of natural population structure in
this branch of A. cerana as does the frag-
mentation of Japan itself into a series of
islands. Two distinct morphoclusters are
currently recognised, one on the islands of
Kyushu, Shikoku and Honshu (bees are not
native to northern Hokkaido) and another
morphocluster occuring only on the small
island of Tsushima in the Straits of Korea
(Fig. 2, area 31a,b). The honeybees of each
of these islands has some unique properties.
For example, for southernmost Kyushu,
Akahira and Sakagami (1959b) demonstrated
a size cline in which the more southerly bees
were larger than their northern counterparts.
8
Infraspecific categories of Apis cerana
and Otis, 1997; Fuchs et al., 1996; Ruttner,
1988).
In another recent study of this region
Damus and Otis (1997) performed multi-
variate analyses of insular Malaysia and
Indonesia and obtained four distinct mor-
phoclusters (Fig. 2, areas 23–26): one iso-
lated island cluster on Timor (area 25) (one
of these in extreme southern Sulawesi in
area 24 was first noted in Hadisoesilo et al.,
1995). The greater part of Indonesia formed
one morphocluster (Fig. 2, area 23) with the
exception of one small cluster in southern
Sulawesi (area 24) and the bees of Sabah,
NE Borneo (Fig. 2, area 26) yet another. In
the classical literature all of these bees
belong to the A. cerana indica complex
(Ruttner, 1988, 1992), but are sometimes
referred to as A. c. javana (Damus and Otis,
1997; Engel, 1999).
Damus and Otis (1997) also included
bees of the Philippines (Fig. 2, areas 28–30)
in their study and concluded that they are
morphometrically distinct from the A. cer-
ana indica of Indonesia. Moreover, they
found that the bees of Luzon were morpho-
metrically distinct from those of Mindanao.
Coupling their morphometric data with the
mtDNA results obtained by Smith and
Hagen (1997) they questioned whether the
bees of Luzon actually belong to any of the
A. cerana groups. We return to this prob-
lem in considering morphometrics, mtDNA
and allozymes conjointly.
Finally, the most recently analysed island
group of honeybees is that of Tilde et al.
(2000) who extensively covered the Philip-
pines using standard multivariate methods.
They found three distinct morphoclusters
(Fig. 2, areas 28–30) corresponding to
Luzon island (with highland and lowland
ecotypes), another morphocluster on the
islands in the Visayas and Mindanao groups
and a quite separate cluster on Palawan. The
bees of Palawan were quite distinct from
the others. All of these bees were tentatively
regarded as A. cerana philippina by Ruttner
(1988).
Likewise, southern bees are lighter in colour
than northern ones (Tokuda, 1924). More-
over, at an interlocality sampling distance
of less than 100 km, intercolonial variance
was low, intracolonial variance high. How-
ever, the intercolonial morphometric homo-
geneity in the variances of honeybees on
Kyushu argues for a fairly uniform single
population with continuous genetic flow
among them.
2.1.3 Southeast Asia
This region extends east of longitude 98°
and southwards from about latitude 20° N to
Timor below the equator at 10° S. The main-
land is about 1.5 million km2(Fig. 2). With
the notable exceptions of Laos, Cambodia
and Vietnam, it is also that area of Asia for
which most of the recent analyses of hon-
eybees have included thorough multivari-
ate statistical analyses as well as analyses
of mitochondrial DNA and various
allozymes (see below).
Moving southwards down peninsular
Indochina, the first study of interest con-
cerns Thailand and Malaysia. Sylvester
et al. (1998) published a comprehensive
morphometric study of the honeybees of
this region and unequivocally established
four distinct morphoclusters (Fig. 2, areas
19–22) one covering most of Thailand, a
second southern Thailand and continental
Malaysia, a third at Phuket island and a
fourth at Samui island. All four of these
morphoclusters could be considered as sub-
sets of what has previously been recognised
as A. cerana indica (Ruttner, 1988, 1992).
Explanations for the distinctness of the
Samui and Phuket morphoclusters have been
reasonably attributed to founder effects
(Sylvester et al., 1998). The meaningfulness
of the designation “A. cerana indica”
(sometimes A. c. javana) for these bees is
put under considerable pressure when it is
remembered that the “cerana indica” of
Thailand, Borneo and Malaysia are certainly
not the same bees called “cerana indica”
which occur in India and Sri Lanka (Damus
9
H.R. Hepburn et al.
2.2. Allozyme diversity
Numerous studies of allozymes in Apis
mellifera have shown relatively little
allozyme diversity in this species (Cornuet
and Garnery, 1991). However such varia-
tion as does occur, particularly in cytoplas-
mic malate dehydrogenase (MDH1, Enzyme
Commission number 1.1.1.37), may have
important metabolic consequences for hon-
eybee flight (Harrison et al., 1996; Hepburn
et al., 1999). MDH1 has proven to be a pow-
erful tool for the investigation of popula-
tion structure and gene flow in A. mellifera
(Meixner et al., 1994; Smith and Glenn,
1994) especially when used in conjunction
with other polymorphic enzymes, such as
non-specific esterases (EST, E. C. number
3.1.1.1) or hexokinase (HK, E. C. number
2.7.1.1). Although allozymes have proved
very useful in studies of A. mellifera, the
study of allozymes in A. cerana is at a very
early stage, and is beset with problems.
Only small portions of the range of
A. cerana have been sampled: Pakistan
(Nunamaker et al., 1984), Sri Lanka
(Sheppard and Berlocher, 1989), Thailand,
peninsular Malaysia, southern Sulawesi and
the Philippines (Gan et al., 1991), Yunnan,
China (Li et al., 1986), Japan (Rozalski
et al., 1996; Tanabe and Tamaki, 1985) and
Korea (Lee and Woo, 1991; Lee et al.,
1989). In addition sample sizes have been
small, on the order of 13 or fewer colonies.
The majority of studies only investigated
variation in MDH1 and/or non-specific
esterases (EST).
Three studies (Gan et al., 1991; Lee and
Woo, 1991; Sheppard and Berlocher, 1989)
carried out a more thorough survey of 10–15
enzyme systems. Although a different suite
of enzymes was surveyed in each study there
is some overlap, particularly between the
Korean (Lee and Woo, 1991) and Sri
Lankan (Sheppard and Berlocher, 1989)
studies. Not surprisingly, studies that sur-
veyed more enzyme systems detected more
variation. All polymorphisms discovered in
these studies consisted of one common allele
present at a frequency of 86% or higher, and
one or more rare alleles. These studies are
summarised in Table I.
Unfortunately, it is not possible to com-
bine data from these studies to examine geo-
graphic patterns of allozyme variation or
draw broader biogeographic conclusions.
In this connection differing buffer systems
are critical for the detection of allozyme
variation and often confound the compar-
isons of results of different studies. Only
the studies by Rozalski et al. (1996) and
Sheppard and Berlocher (1989) provided a
standardised nomenclature of alleles. In their
studies, putative alleles of A. cerana were
compared to the alleles found in A. mellifera,
and named according to their relative elec-
trophoretic mobility. Another useful prac-
tice followed by these authors was to include
A. mellifera “standards” on gels, that is,
samples with known genotypes.
Tentative among-region comparisons can
be made for MDH and EST, the two
enzymes most commonly surveyed. Japan
and Pakistan samples showed only a single
MDH1 allele, while samples from all other
locations showed two alleles. In Korea and
Sri Lanka, the “fast” allele was reported to
be more common than the “slow” allele (this
information was not provided for the Thai,
Malay, Indonesian and Philippine samples).
It is possible that MHD1107 from Japan,
MHD1109 from Sri Lanka, and MDH1fast
from Korea and Thai, Malay, Indonesian
and Philippine samples all correspond to
the same common electromorph, while
MHD175 from Sri Lanka, and MDH1slow
from Korea and Thai, Malay, Indonesian
and Philippine samples all correspond to
the same rarer allele, but this cannot be con-
firmed without more direct comparisons.
Some evidence of geographic variation in
allele frequency is apparent in EST. This
enzyme was reported to be monomorphic
in Pakistan, China and Korea. Japan, Sri
Lanka, and the Thai, Malay, Indonesian and
Philippine samples each showed two
10
Infraspecific categories of Apis cerana 11
Table I. Summary of allozyme studies of Apis cerana (where alleles are named according to their relative mobility, authors used A. mellifera standards).*
Locality Sample size Enzymes Polymorphism Alleles Frequency Reference
Rawalpindi, 12 colonies, EST No Nunamaker et al., 1984
Pakistan 100 bees total MDH1 No
Meng La, 100 bees EST No
southwest Yunnan, Li et al., 1986
China
Japan, 13 colonies, EST Yes EST73 0.96a
9 locations 12–48 bees/colony, EST63 0.04 Rozalski et al., 1996
405 workers, 48 drones MDH1 No MDH1107
Korea 5 colonies, EST No MDH1fast 0.86 Lee et al., 1989
27–30 bees/colony MDH1 Yes MDH1slow 0.14
Korea 5 apiaries, MDH1 Yes MDH1fast Lee and Woo, 1991
20–41 bees/apiary MDH1slow
ACPH,APH, EST, α-GPDH,
HK, IDH, ME, ODH, PGM No
& XDH all monomorphic
Sri Lanka, 10 colonies, ACON2 Yes ACON2100 0.97–1.00b
10 locations ≥15 workers/colony ACON2114 0.03–0.00 Sheppard and Berlocher, 1989
EST Yes EST86 0.03–1.00
EST57 0.97–1.00
ME ME110 0.03–0.00
ME91 0.97–1.00
MDH1 MDH1109 0.95–1.00
H.R. Hepburn et al.
12
ACON1, ALDO, ARGK, MDH175 0.05–0.00
G-3-PDH, α-GPDH, No
β-HBDH, HK, IDH, LAP,
PGM & TPI all monomorphic
Bangkok, Thailand unspecified EST No EST70 most common Gan et al., 1991
Peninsular Malaysia EST100 least common
Sabah, Borneo FUM Yes 1 common 4 rare alleles
south Sulawesi α-GPDH Yes 1 common 2 rare alleles
Indonesia GLDH Yes 2 common 1 rare allele
Luzon, Philippines MDH Yes 2 alleles
SUDH Yes 1 common 1 or 2 rare alleles
APH, ACPH, 6-PGD
& SHDH all monomorphic
* Enzymes mentioned in text, with abbreviation and enzyme commission numbers in parentheses: ACON = aconitase (4.2.1.2); ACPH = acid phosphatase (3.1.3.2);
ALDO = aldolase (4.1.2.13); APH = alkaline phosphatase (3.1.3.1); ARGK = arginine kinase (3.3.8.9); EST = non-specific esterase (3.1.1.1); FUM = fumarase (4.2.1.2);
G-3-PDH = glyceraldehyde-3-phosphate dehydrogenase (1.2.1.12); GLDH = glucose dehydrogenase (1.1.1.47); a-GPDH = α-glycerophosphate dehydrogenase (1.1.1.8);
β-HBDH = β-hydroxybutyric acid dehydrogenase (1.1.1.30); HK = hexokinase (2.7.1.1); IDH = isocitrate dehydrogenase (1.1.1.42); LAP = leucine amino peptidase
(3.4.1.1); MDH1 = cytoplasmic malate dehydrogenase (1.1.1.37); ME = malate dehydrogenase (1.1.1.40); ODH = octanol dehydrogenase (1.1.1.73); 6-PGD = 6-phos-
phogluconate dehydrogenase (1.1.1.43,44); PGI = phosphoglucose isomerase (5.3.1.9); PGM = phosphoglucomutase (2.7.5.1); SHDH = shikimate dehydrogenase (1.1.1.25);
SUDH = succinate dehydrogenase (1.3.99.1); TPI = triose phosphate isomerase (5.3.1.1); XDH = xanthine dehydrogenase (1.2.1.37).
aGenotypes of queens were inferred from worker and drone genotypes, allele frequencies estimated from 13 queen genotypes.
bRange of allele frequencies found in colonies.
Table I. Continued.
Locality Sample size Enzymes Polymorphism Alleles Frequency Reference
Infraspecific categories of Apis cerana
Philippines: de la Rúa et al., 2000; Smith
et al., 2000). These studies have used several
techniques for detection of variation. The
earliest study surveyed restriction enzyme
cleavage sites over the entire mitochondrial
genome of A. cerana samples (Smith, 1991),
while more recent studies PCR-amplify frag-
ments of the mitochondrial genome, and
screen for variation in restriction enzyme
cleavage sites or sequence (Arias et al.,
1996).
All of the more recent studies have
focused on one region of the honey bee
mitochondrial genome, from the cytochrome
oxidase I gene (COI) to the cytochrome oxi-
dase II gene (COII). Between COI and COII,
lie the leucine tRNAUUR gene and a non-
coding sequence that is apparently unique
to Apis (Cornuet et al., 1991). The non-cod-
ing region is small in A. florea, A. andreni-
formis and A. dorsata (on the order of
24–32 bases), but larger in the cavity-nest-
ing bees (89 to 97 in A. cerana, 94 in
A. koschevnikovi, ~200–900 in A. mellif-
era). Because it is non-coding, this sequence
is free to evolve rapidly, and provides infor-
mation analysable at the intraspecific level.
In addition to this region, Sihanuntavong
et al. (1999) also examined PCR-amplified
fragments containing portions of the genes
for the small and large subunit ribosomal
RNA genes (ssRNA and lsRNA).
Comparative, macroscale studies have
employed both restriction fragment length
polymorphisms (de la Rúa et al., 2000;
Deowanish et al., 1996; Smith, 1991) and
DNA sequence of the non-coding region
(de la Rúa et al., 2000; Smith and Hagen,
1997, 1999; Smith et al., 2000) and there is
only partial overlap among these studies in
the geographic regions sampled. Nonethe-
less, results of these studies are largely con-
gruent. Groups detected by all comparative
studies are: (1) mainland Asia including
Japan; (2) Sundaland (including southern
or peninsular Thailand and the island of
Samui); (3) Palawan (Philippines); and (4)
the oceanic islands of the Philippines (Fig. 3).
alleles. In Japan, the fast allele (EST73) was
most common, while in the other locations
the slow allele (EST57 in Sri Lanka, EST70 in
Thai, Malay, Indonesian and Philippine)
was most common. This may indicate a gen-
eral difference between northern and south-
ern A. cerana populations, although the issue
is complicated by the possibility that there
are multiple loci for non-specific esterases in
A. cerana (e.g. Gan et al., 1991).
A simple comparison of P, the propor-
tion of polymorphic loci, among sites is dif-
ficult. Most studies only examined Mdh and
Est, and the three studies that examined
more loci did not examine the same set of
enzymes. Empirically, some enzymes (e.g.
Mdh1, and phosphoglucomutase, Pgm) fre-
quently show variation, while other enzymes
are less likely to show variation, so that the
proportion of polymorphic loci will be
biased by the set of enzymes surveyed. At
present insufficient information on enzyme
polymorphism (or lack thereof) is available
from which to make any major inferences
about geographic variation in allele fre-
quencies in A. cerana.
2.3. Mitochondrial DNA diversity
MtDNA haplotypes have proven to be
an incisive tool for unravelling the popula-
tion structure of A. mellifera (Hall and
Smith, 1991; Smith et al., 1991). By com-
parison, studies of the mtDNA of A. cerana
and other Asian honeybees are in their
infancy, even though A. cerana occupies an
area comparable in size to that of A. mellif-
era. To date, nine studies have been pub-
lished on the mtDNA of A. cerana. Most of
these are comparative, surveying samples
from numerous geographic locations but
with relatively few samples per location (de
la Rúa et al., 2000; Deowanish et al., 1996;
Smith, 1991; Smith and Hagen, 1997, 1999;
Smith et al., 2000). Some provide a more
intensive survey of variation in a particular
geographic location (Thailand: Deowanish
et al., 1996; Sihanuntavong et al., 1999;
13
H.R. Hepburn et al.
The Asian mainland group contains a
large number of related haplotypes. It
includes samples from Japan, Korea, China
(Hong Kong, Yunnan), Nepal, Vietnam
(northern and southern), northern Thailand,
and some of the bees from India. Deowan-
ish et al. (1996) were also able to discrimi-
nate samples from Honshu Island, Japan,
from their other mainland Asian samples
by a HaeIII restriction site polymorphism
in a fragment of mtDNA including part of
the leucine tRNA gene, the non-coding
region, and the 5’ end of the COII gene.
Studies using the sequence of the non-cod-
ing region alone (Smith and Hagen, 1997,
1999; Smith et al., 2000) were unable to dis-
criminate Japanese bees from the bees of
Korea and other parts of the mainland.
The Sundaland group of haplotypes
includes samples from peninsular Thailand
14
Figure 3. Geographical distribution of major mtDNA groups and subgroups for A. cerana. Stars
indicate areas of high mtDNA diversity. MtDNA groups are indicated as follows: group 1 consists of
mainland Asia with subgroups 1a = Japan, 1b = southern India, 1c = Himalayan region, Indochina
peninsula, southeastern China and Korea; group 2 consists of Sundaland region of peninsula Thailand,
Malaysia and Indonesia; group 3 comprises Palawan (Philippines) and group 4 comprises the oceanic
islands of the Philippines. Areas designated with a question mark do not yet show sufficiently clear
affinities to assign them to any of the four groups recognised on the map. Original data: Arias and
Sheppard, 1996; Arias et al., 1996; de la Rúa et al., 2000; Deowanish et al., 1996; Sihanuntavong
et al., 1999; Smith, 1991; Smith and Hagen, 1997, 1999; Smith et al., 2000.
Infraspecific categories of Apis cerana
(Deowanish et al., 1996; Smith and Hagen,
1997, 1999; Smith et al., 2000), and with
morphometric data (Limbipichai, 1990;
Sylvester et al., 1998). Thus the northern
and central Thai samples belong to the Asian
mainland group, while the southern Thai,
Samui and probably the Phuket samples
belong to the Sundaland group.
The shift from Asian mainland to Sun-
daland haplotypes occurs in the Isthmus of
Kra at the so-called Kra ecotone, where there
is a transition from evergreen rainforest
(south of the imaginary line joining the cities
of Kangar and Pattani) to more seasonal,
semi-evergreen forest (Whitmore, 1984).
The Bilauktaung mountain range, which
forms part of the boundary between Myan-
mar (Burma) and Thailand in the Malay
peninsula, may also provide some impedi-
ment to gene flow between Mainland and
Sundaland populations.
The islands of the Philippines are home to
a diverse collection of A. cerana popula-
tions, belonging to at least two mitochon-
drial lineages: the Palawan group and the
oceanic Philippine islands group (Fig. 3).
Mesoscale studies of Philippine bees from
the islands of Palawan, Luzon, Mindanao,
Panay, Negros, Cebu and Leyte were car-
ried out by de la Rúa et al. (2000) and Smith
et al. (2000). Both employed a set of colony
samples (47 and 29 colonies, respectively)
provided by the Bee Program of the Uni-
versity of the Philippines, Los Banos; Smith
et al. (2000) also employed an additional
10 samples provided by other collectors.
Palawan, Luzon and Mindanao had the best
coverage, while Negros, Panay, Leyte and
Cebu were each represented by 1–3 colonies.
Both groups PCR-amplified a region of the
mitochondrial genome that included the
leucine tRNA gene, the non-coding region
and the 5’ end of COII. The approach used
by de la Rúa et al. (2000) was to digest the
amplification products with the restriction
enzyme DraI to detect variation, and then
sequence exemplars of each restriction pat-
tern. This approach enables rapid screening
and Malaysia, and the islands of Samui,
Phuket, Borneo, Java, Bali, Lombok, Timor
and Flores as well as S. Sulawesi. The
islands (Fig. 3, area 2) lie on the broad
Sunda continental shelf of southeast Asia
(Heaney, 1985, 1986, 1991). During Pleis-
tocene episodes of glaciation, water accu-
mulated in glaciers and polar icecaps, low-
ering global sea levels. During the
mid-Pleistocene glaciation (160000 years
ago), sea levels were approximately 160 m
lower than at present, and during the late
Pleistocene (16000 to 18 000 years ago)
120 m lower (Heaney and Rickart, 1990).
During periods of low sea level, the islands
on the Sunda shelf would have been united
directly with the Asian mainland by dry
land, forming the region known as Sunda-
land. Lombok, Flores and Timor would have
been separated from the larger landmass by
narrow channels. This would have made
possible migration and gene flow between
continental Asia and Sundaland, followed
by isolation of Sundaland populations on
islands as sea levels rose.
Sihanuntavong et al. (1999) provide an
extensive mesoscale analysis of mtDNA
diversity in Thailand based on 170 colonies
of honeybees from some 122 localities
throughout Thailand. They PCR-amplified
three regions of the mitochondrial genome
(the COI to COII region, ssRNA and
lsRNA) and digested the resulting fragments
with the 6 base restriction enzyme DraI.
They calculated the genetic distance between
composite mtDNA haplotypes, the haplo-
type and nucleotide diversity within sam-
ples and nucleotide divergence between
samples as well as the genetic heterogeneity
between geographic samples. A total of
12 composite haplotypes was observed
(Sihanuntavong et al., 1999). The geo-
graphic distribution of haplotypes indicated
a genetic discontinuity between the bees of
northern and peninsular Thailand (Fig. 3,
areas 1 and 2a). The bees of Samui island
formed a distinct group but those of Phuket
island did not. These results are consistent
with comparative studies discussed above
15
H.R. Hepburn et al.
of large numbers of colonies. Using this
method they detected four haplotypes: two
on Palawan, one on Luzon, and one on the
remaining islands. Smith et al. (2000)
sequenced the non-coding region of each
sample; this is slower, but maximises the
variation detected. Using this approach,
11 haplotypes were detected. The broad
results of the two studies are congruent,
showing haplotype frequency differences
between three regions: Palawan, Luzon, and
Mindanao plus the other islands. The stud-
ies differ in the amount of variation detected,
particularly in Panay, Negros, Cebu and
Leyte.
The distributions and relationships of
these haplotypes can also be interpreted in
light of changes in sea level during the Pleis-
tocene (Heaney, 1985, 1986, 1991; Heaney
and Rickart, 1990). During low sea level
periods, islands of the Philippines were
joined into larger units, or “mega islands”:
Greater Palawan, Greater Luzon, Greater
Mindanao (which includes modern Min-
danao, Samar, Leyte, and Bohol), and
Greater Negros-Panay (which includes mod-
ern Negros, Panay, Cebu and Masbate).
Greater Luzon and Greater Mindanao may
have been joined in the mid-Pleistocene,
though this is not certain. (Today the trench
separating Luzon and Samar is only 140 m
deep, but this is currently a region of geo-
logical uplift, so the trench may have been
deeper during the Pleistocene.)
Today Palawan is separated from nearby
Borneo by a 145 m trench; thus Greater
Palawan would have been united with the
Asian mainland through Borneo in the mid
Pleistocene, when sea levels were 160 m
lower than present, but not in the late Pleis-
tocene, when sea levels were only 120 m
lower than present. The Palawan haplotype
group includes two haplotypes found only
on Palawan, and two found on Panay and
northwestern Mindanao. Neighbour-join-
ing and parsimony analyses (Smith et al.,
2000) show the Palawan sequences more
closely related to Sundaland and mainland
Asian sequences than to other sequences
from the Philippines.
The other Philippine islands have had no
above-water connection to the Asian main-
land, though two island chains, the Palawan
chain between Borneo and Mindoro, and
the Sulu Archipelago between Borneo and
Mindanao, may have formed “stepping
stones” between Borneo and the oceanic
islands of the Philippines. The haplotypes
on these islands appear to be the most diver-
gent from other A. cerana haplotypes.
A current working hypothesis on popu-
lation structure based on variability of
mtDNA haplotypes is shown in Figure 3.
Combining all of the currently available data
it would appear there are four major groups
of mtDNA lineages. These are (1) an Asian
mainland mtDNA lineage, and three addi-
tional mtDNA lineages occurring on lands
connected to the mainland for successively
shorter periods of time: (2) Sundaland (con-
nected to the mainland during the mid and
late Pleistocene), (3) Palawan (connected
only during the mid-Pleistocene), and (4)
the oceanic islands of the Philippines (never
connected to the mainland). The Indonesian
island of Sulawesi too, was never connected
to the mainland. On Sulawesi we find two
exceedingly similar cavity-nesting bee
species, A. nigrocincta and A. cerana with
Sundaland mtDNA. The affinities of the
“yellow” or “plains” bees of India and Sri
Lanka are still uncertain, as are those of hap-
lotypes lacking most of the non-coding
sequence (all samples from Taiwan, and
some from Sulawesi and the Philippines).
Discrete groups can be recognised within
some of the major groups of mtDNA lin-
eages. Within the large Asian mainland
group, some haplotypes from “black” or
“hill” bees form a distinct group; some
authors have also been able to discriminate
the bees of Japan from the other mainland
samples. Within the Sundaland group, sam-
ples from Samui, peninsular Malaysia, and
Borneo form distinct clusters (Fig. 3, areas
a–c).
16
Infraspecific categories of Apis cerana
This state of affairs is worrying for a
number of reasons. As examples, note that
the ecotypical studies of the honeybees of
the Indian subcontinent reported by Kshir-
sagar (1983) cannot be rationalised with the
multivariate studies of Singh et al. (1990)
or of Verma et al. (1994) because of differ-
ent methods of analysis, completely differ-
ent databases and the absence of the raw
databases. Similarly, the thorough and pre-
cise studies of Singh et al. (1990) on the
honeybees of the western Himalayas can-
not be added to the similarly precise work of
Verma et al. (1994) on eastern Himalayan
bees because of fundamentally different
databases and the absence of sufficient raw
data to provide for new and combined anal-
yses. Similarly, equally thorough studies
touching on Malaysia, Indonesia and
the Philippines (Damus and Otis, 1997;
Sylvester et al., 1998; Tilde et al., 2000)
cannot be amalgamated into a single syn-
thesis for this southern Asian region. More-
over, there remain as yet unsampled regions
of A. cerana distribution which are quite
considerably greater in extent than those
that have been sampled.
Numerous anomalies become particularly
apparent if a composite of the geographical
distribution of all putative subspecies of
A. cerana is attempted as in Figure 4. For
example, the apparent disjunct distribution
for A. c. cerana (separated by A. c. sko-
rikovi) is unlikely to prove real given
detailed analyses. Similarly the reality of
A. c. skorikovi and A. c. abaensis derives
from one set of authors (Peng et al., 1989)
but is not considered by others (Ruttner,
1988, 1992). Moreover, it is readily appar-
ent that most recent authors clearly distin-
guish minimally two different sets of what is
currently termed A. c. indica, one to the west
and the other eastern in distribution, the lat-
ter sometimes being called “javana” (Fig. 4).
Clarification of this anomaly can only be
achieved when there is a sufficient database
for Myanmar, Laos, Cambodia and Viet-
nam. Many other more localised anomalies
are apparent.
3. CONCLUDING REMARKS
Although a relatively large amount of
information on the infraspecific classifica-
tion of A. cerana is available, it varies quite
considerably in quality and kind. Under-
standably, it ranges from the anecdotally
descriptive accounts of earlier decades
through univariate methods of analysis and
finally to the more recent application of
complete multivariate analyses of morpho-
metric characters for specific regions. The
greatest obstacle to a reasoned synthesis of
infraspecific categories in A. cerana at pre-
sent is that the results of most of such stud-
ies cannot be collated and unified because of
fundamental and incompatible differences
in statistical analyses, sample sizes, char-
acter suites, morphocluster confidence lim-
its, the critical elements of sampling dis-
tance and extent of geographical coverage
(Daly, 1991, 1992; Ruttner, 1988; Ruttner
et al., 1978). Recent studies of A. mellifera
show that morphocluster formation and
inclusivity (correct classification) are highly
sensitive to sampling distance intervals.
Indeed, the length of a transect may obscure
small biometric groups when the between-
group variation is considerably larger than
the within-group variation. Likewise, vary-
ing the limits of confidence applied to the
ellipses and the discriminant a posteriori
probabilities from low to high also decreases
the numbers of colonies correctly assignable
to morphoclusters (Diniz-Filho et al., 2000;
Hepburn et al., 2000; Radloff and Hepburn,
1998). In the circumstances, specific rec-
ommendations for an approach to future
morphometric analyses of A. cerana
populations are currently being prepared
(Hepburn et al., in preparation). Resolution
of this problem is further exacerbated by
the fact that the relevant literature largely
ignores the recommendations of the Inter-
national Code of Zoological Nomenclature
(Engel, 1999). Nonetheless it is too prema-
ture to apply such rules until such time as
the groups of cerana bees can be treated in
a single, unified analysis.
17
H.R. Hepburn et al.
The putative morphometric portrait of
A. cerana shown in Figure 2 with a classi-
fication structure as in Figure 4 cannot be
considered a reality as of yet. Extraordinary
gaps must be filled for Afghanistan, Pak-
istan, west and central India, Myanmar,
Laos, Cambodia, Vietnam, southern and
northeastern China and Indonesia, the
Andaman and Nicobar islands and perhaps
Mongolia. Given these uncertainties, a com-
parison of morphometric results with those
of mtDNA are equally disconcerting. There
is a reasonable geographic congruity only
between the two character sets for honey-
bees of the island systems Japan, Philip-
pines, Taiwan, Phuket and Samui in Thai-
land and for peninsular Malaysia and Korea.
All other morphoclusters and mtDNA clus-
ters show little or no geographical corre-
spondence. Turning to enzyme polymor-
phism (Fig. 3), the available data, as for
mtDNA, is extremely sparse. In any event,
it appears that low enzyme polymorphism
occurs towards the two end points of A. cer-
ana distribution (e.g. northeastern China
and northern India). Conversely, high
enzyme polymorphism occurs among the
island systems (Sri Lanka, Sulawesi and the
Philippines) and the Indochina peninsula.
The massive continental mainland remains
completely unexplored in this regard.
18
Figure 4. Composite geographical distribution of putative subspecies of A. cerana. Numerous unre-
solved anomalies are readily apparent and are discussed in the text. References as for Figure 2.
Infraspecific categories of Apis cerana
et portent uniquement sur le gène de la cyto-
chrome oxydase. Néanmoins plusieurs
groupes distincts d’haplotypes émergent des
études comparatives : (i) l’Asie continen-
tale et le Japon, (ii) le Sundaland (péninsule
malaysienne + îles de la Sonde, Bornéo) y
compris la Thaïlande, (iii) Palawan (îles
Philippines) et (iv) les îles océaniques des
Philippines (Fig. 3). La répartition de ces
groupes peut s’interpréter en relation avec
les variations du niveau des mers au Pléis-
tocène.
À cause d’incompatibilités méthodologiques
entre les études morphométriques, allozy-
miques et d’ADNmt et de lacunes impor-
tantes dans l’échantillonnage géographique,
il n’est pas encore possible d’établir des
catégories infraspécifiques significatives
pour A. cerana. De même, il n’est pas encore
possible de faire des déductions importantes
concernant les catégories infraspécifiques
en combinant les données morphométriques,
allozymiques et d’ADNmt, ni de fournir des
dénominations acceptables du point de vue
de la nomenclature pour les sous espèces
d’A. cerana.
Apis cerana / taxonomie / biogéographie /
Asie / abeilles
Zusammenfassung – Innerartliche Ein-
ordnung von Apis cerana: Morphometrie
und Unterschiede bei Allozymen und mt
DNA. Die natürliche Verbreitung von Apis
cerana ist gröβtenteils kontinental (Abb. 1).
Analysen von publizierten Arbeiten über
Taxonomie und Morphometrie dieser Art
enthalten 31 mutmaβliche biometrische
Gruppen (Abb. 2) oder vielleicht 8 Unter-
arten mit verschiedenen Ökotypen (Abb. 4).
Diese Ergebnisse sind durch eine Zusam-
menfassung verschiedener Untersuchungen
entstanden, die von anekdotischen Bewer-
tungen bis zu multivariaten Analysen rei-
chen. Entsprechend ist eine vollständige
Synthese von morphometrisch definierten
innerartlichen Kategorien im Moment nicht
möglich, vor allem wegen der fundamentalen
ACKNOWLEDGEMENTS
We thank P. de la Rúa, P. Neumann, Choon
Thin Yat and S. Wongsiri for their help and
advice with this manuscript. We especially thank
P. Munn of IBRA for providing relevant litera-
ture.
Résumé – Les catégories infraspécifiques
d’Apis cerana : diversité morphologique,
allozymique et de l’ADNmt. La réparti-
tion actuelle d’A. cerana est continentale
(Fig. 1). L’analyse des travaux publiés sur la
taxonomie et la morphologie de cette espèce
fournit environ 31 groupes biométriques
possibles (Fig. 2) ou peut-être huit sous-
espèces avec plusieurs écotypes (Fig. 4).
Ces résultats proviennent d’un amalgame
d’études allant de l’analyse anecdotique à
l’analyse multivariée. Aussi une synthèse
complète des catégories infraspécifiques
définies par la morphométrie est elle actuel-
lement exclue en raison de différences fon-
damentales et incompatibles entre les études
quant à la taille de l’échantillon, les choix
des caractères, la couverture géographique,
la distance d’échantillonnage, les limites de
confiance, la méthodologie statistique et
l’absence courante d’une base de données
continentale unifiée.
Les études de polymorphisme allozymique
portant sur A. cerana sont très récentes mais
elles rencontrent des problèmes. Peu de
régions ont été étudiées, les échantillons
sont de petite taille et le nombre de systèmes
enzymatiques analysés est limité ; il s’agit
principalement de la malate déshydrogé-
nase et d’estérases non spécifiques. Mais
l’étendue de la variation est liée au nombre
de système enzymatiques testés. Les résul-
tats des diverses études ne peuvent pas non
plus être combinés car il n’existe aucune
nomenclature normalisée des allèles. Il est
donc difficile de déterminer les proportions
de locus polymorphiques, ce qui exclut
toutes déductions importantes concernant
les catégories infraspécifiques.
Les analyses des haplotypes d’ADNmt sont
peu nombreuses, limitées en échantillonnage
19
H.R. Hepburn et al.
Unterschiede und Widersprüche zwischen
den Studien in Bezug auf Probengröβe,
Zusammensetzung der Merkmale, geogra-
phische Herkünfte, Abstände bei der Pro-
bennahme, Vertrauensgrenzen, statistische
Methoden und auf Grund des momentanen
Fehlens einer vereinheitlichten kontinenta-
len Datengrundlage.
Die Untersuchungen über Allozyme von
Apis cerana sind zwar erst vor kurzem
durchgeführt worden, aber auch sie erweisen
sich als problematisch. Nur wenige Regio-
nen sind bisher untersucht worden, die Zahl
der Proben ist klein und der Bereich der ana-
lysierten Enzymsysteme ist vor allem auf
Malatdehydrogenase und unspezifische
Esterasen begrenzt. Das Ausmaβder Varia-
tion aber hängt von der Anzahl der unter-
suchten Enzymsysteme ab. Auch diese
Ergebnisse der verschiedenen Untersu-
chungen können nicht kombiniert werden.
Es gibt keine standardisierte Nomenklatur
der Allele. So ist es schwierig, den Anteil
der polymorphen Loci zu bestimmen und
dadurch können keine gröβeren Rück-
schlüsse über innerartliche Kategorien
gemacht werden.
Analysen über mtDNA gibt es nur wenige,
die Proben sind limitiert und nur auf das
Gen der Cytochromoxidase konzentriert.
Nichtsdestotrotz ergeben sich mehrere
distinkte Gruppen aus diesen vergleichenden
Arbeiten: 1. Festland Asien und Japan,
2. Sundaland inklusive Thailand, 3. Pala-
wan (Philippinen) und 4. die Ozeaninseln
der Philippinen (Abb. 3). Die Verteilung
dieser Gruppen kann in Bezug auf die Ände-
rungen des Meeresspiegels im Pleistozen
interpretiert werden.
Auf Grund der nicht vergleichbaren Metho-
den bei der Morphometrie, den Allozymen
und den mtDNA Studien und den groβen
geographischen Lücken bei den Proben ist es
noch nicht möglich, sinnvolle innerartlichen
Kategorien von Apis cerana für eine Ablei-
tung der Abstammung aufzustellen. Ent-
sprechend ist es zur Zeit weder möglich,
wichtige Schlüsse über die innerartliche
Einordnung aus einer Kombination von
Morphometrie, Allozymen und mtDNA zu
ziehen noch eine nomenklatorisch akzepta-
ble Benennung für die Unterarten der Apis
cerana vorzunehmen.
Apis cerana / Taxonomie / Biogeographie /
Asien / Honigbienen
REFERENCES
Akahira Y., Sakagami S.F. (1959a) Notes of the dif-
ference in some external characteristics between
Japanese (A. cerana cerana) and European honey-
bees (A. m. ligustica), Annotat. Zool. Jap. 32, 35–42.
Akahira Y., Sakagami S.F. (1959b) A biometrical study
on the Japanese honey bee, observations upon some
populations of Kyushu, J. Hokkaido Gakugei Univ.
14, 175–184.
Arias M.C., Sheppard W.S. (1996) Molecular phylo-
genetics of honey bee subspecies (Apis mellifera
L.) inferred from mitochondrial DNA sequence,
Mol. Phylogenet. Evol. 5, 557–566.
Arias M.C., Tingek S., Kelitu A., Sheppard W.S. (1996)
Apis nuluensis Tingek, Koeniger and Koeniger,
1996 and its genetic relationship with sympatric
species inferred from DNA sequences, Apidologie
27, 415–422.
Avetisyan G.A. (1960) On certain problems of evolu-
tion, geographical distribution, protection and util-
isation of species and races of bees, XVII Congr.
Int. Apic., pp. 223–229.
Cornuet J.M., Garnery L. (1991) Genetic diversity in
Apis mellifera, in: Smith D. (Ed.), Diversity in the
Genus Apis, Westview, Boulder, pp. 103–115.
Cornuet J.M., Garnery L., Solignac M. (1991) Puta-
tive origin and function of the intergenic region
between COI and COII of Apis mellifera L. mito-
chondrial DNA, Genetics 128, 393–403.
Daly H.V. (1991) Systematics and identification of
Africanized honey bees, in: Spivak M., Fletcher
D.J.C., Breed M.D. (Eds.), The “African” Honey
Bee, Westview, Boulder, pp. 13–44.
Daly H.V. (1992) A statistical and empirical evaluation
of some morphometric variables of honey bee clas-
sification, in: Sorensen J.T., Footit R.J. (Eds.), Ordi-
nation in the Study of Morphology, Evolution and
Systematics of Insects: Applications and Quanti-
tative Genetic Rationals, Elsevier, Amsterdam,
pp. 127–155.
Damus M.S., Otis G.W. (1997) A morphometric anal-
ysis of Apis cerana F. and Apis nigrocincta Smith
populations from southeast Asia, Apidologie 28,
309–323.
de la Rúa P., Simon U.E., Tilde A.C., Moritz R.F.A.,
Fuchs S. (2000) mtDNA variations in Apis cerana
populations from the Philippines, Heredity 84,
124–130.
20
Infraspecific categories of Apis cerana
Hepburn H.R., Radloff S.E., Fuchs S. (1999) Flight
machinery dimensions of honeybees, Apis mellifera,
J. Comp. Physiol. B 169, 107–112.
Hepburn H.R., Radloff S.E., Oghiake S. (2000) Moun-
tain honeybees of Africa, Apidologie 31, 205–221.
Kapil R.P. (1956) Variation in biometrical characters
of the Indian honey-bee (Apis indica F.), Indian
J. Entomol. 18, 440–457.
Kshirsagar K.K. (1973) Comparative biometric studies
on Indian honeybees. III. Preliminary observations
on biometry of Apis cerana F. queen, Indian Bee
J. 35, 21–26.
Kshirsagar K.K. (1983) Morphometric studies on the
Indian hive bee Apis cerana indica F. I-morpho-
metric characters useful in identification of intraspe-
cific taxa, Proc. 2nd Int. Conf. Apic. Trop. Clim.,
New Delhi, pp. 254–261.
Kwon Y.J., Huh E.Y. (1992) Electron-morphometric
classification of the native honeybees from Korea,
Part VI. Cluster analysis by arithmetic mean,
Korean J. Apic. 7, 118–129.
Lawrjochin F.A. (1960) Über die ersten Versuche zur
Einführung der wildlebenden ussurischen Bienen
(Apis indica F.) in den europäischen Teil der Sow-
jetunion, XVII Congr. Int. Apic., pp. 237–238.
Lee M.L., Woo K.S. (1991) Enzyme polymorphism
of Apis cerana Fabr. in Korea, Honeybee Sci. 12,
58–60 (in Japanese).
Lee M.L., Yim Y.H., Kim S.S., Woo K.S., Suh D.S.
(1989) Malate dehydrogenase and non-specific
esterase polymorphism in Apis mellifera L. and
Apis cerena F. in South Korea, Korean J. Apic. 4,
68–74 (in Korean).
Li S., Meng J.T., Chang J., Li S., He S., Kuang B.
(1986) A comparative study of esterase isozymes in
6 species of Apis and 9 genera of Apoidea, J. Apic.
Res. 25, 129–133.
Limbipichai K. (1990) Morphometric studies on the
eastern honey bee (Apis cerana Fabricius) in Thai-
land and the Malaysian peninsula, Thesis, Chula-
longkorn University, Bangkok.
Maa T. (1953) An inquiry into the systematics of the
Tribus Apidini or honeybees (Hymenoptera),
Treubia 21, 525–640.
Mattu V.K., Verma L.R. (1983a) Comparative mor-
phometric studies on introduced European bee Apis
mellifera L. and Indian honeybee Apis cerana
indica F. in Himachal Pradesh, Proc. 2nd Int. Conf.
Apic. Trop. Clim., New Delhi, pp. 262–277.
Mattu V.K., Verma L.R. (1983b) Comparative mor-
phometric studies on the Indian honeybee of the
north-west Himalayas. 1. Tongue and antenna,
J. Apic. Res. 22, 79–85.
Mattu V.K., Verma L.R. (1984a) Comparative mor-
phometric studies on the Indian honeybee of the
north-west Himalayas. 2. Wings, J. Apic. Res. 23,
3–10.
Mattu V.K., Verma L.R. (1984b) Comparative mor-
phometric studies on the Indian honeybee of the
north-west Himalayas. 3. Hind leg, tergites and
sternites, J. Apic. Res. 23, 117–122.
Deowanish S., Nakamura J., Matsuka M., Kimura K.
(1996) mtDNA variation among subspecies of Apis
cerana using restriction fragment length polymor-
phism, Apidologie 27, 407–413.
Diniz-Filho J.A.F., Malaspina O., Pignata M.I.B. (1993)
Geographic variation in Apis cerana indica F.:
a spatial autocorrelation analysis of morphometric
patterns, J. Apic. Res. 32, 65–72.
Diniz-Filho J.A.F., Hepburn H.R., Radloff S.E., Fuchs
S. (2000) Spatial analysis of morphological varia-
tion in African honeybees (Apis mellifera L.) on a
continental scale, Apidologie 31, 191–204.
DuPraw E.J. (1964) Non-Linnean taxonomy, Nature
202, 849–852.
DuPraw E.J. (1965) Non-Linnean taxonomy and the
systematics of honeybees, Syst. Zool. 14, 1–24.
Engel M.S. (1999) The taxonomy of recent and fossil
honeybees (Hymenoptera: Apidae; Apis), J. Hym.
Res. 8, 165–196.
Fernando E.F.W. (1979) Some biometrical features of
Apis cerana F. from Sri Lanka, Indian Bee J. 41,
5–8.
Fuchs S., Koeniger N., Tingek S. (1996) The morpho-
metric position of Apis nuluensis within cavity-
nesting honeybees, Apidologie 27, 397–405.
Gan Y.Y., Otis G.W., Mardan M., Tan S.G. (1991)
Allozyme diversity in Asian Apis, in: Smith D.
(Ed.), Diversity in the Genus Apis, Westview, Boul-
der, pp. 117–130.
Hadisoesilo S., Otis G.W., Meixner M. (1995) Two
distinct populations of cavity-nesting honey bees
(Hymenoptera: Apidae) in South Sulawesi, Indone-
sia, J. Kans. Entomol. Soc. 68, 399–407.
Hall H.G., Smith D.R. (1991) Distinguishing African
and European honeybee matrilines using amplified
mitochondrial DNA, Proc. Nat. Acad. Sci. USA
88, 4548–4552.
Harrison J.F., Nielsen D.I., Page R.E. (1996) Malate
dehydrogenase phenotype, temperature and colony
effects on flight metabolic rate in the honey-bee,
Apis mellifera, Funct. Ecol. 10, 81–88.
Heaney L.R. (1985) Zoogeographic evidence for mid-
dle and late Pleistocene land bridges to the Philip-
pine Islands, Mod. Quarternary Res. SE Asia 9,
127–143.
Heaney L.R. (1986) Biogeography of mammals in SE
Asia: estimates of rates of colonization, extinction
and speciation, Biol. J. Linnaean Soc. 28, 127–165.
Heaney L.R. (1991) A synopsis of climate and vege-
tational change in southeast Asia, Climate Change
19, 53–61.
Heaney L.R., Rickart E.A. (1990) Correlations of clades
and clines: geographic, elevational and phyloge-
netic distribution patterns among Philippine mam-
mals, in: Peters G., Hutterer R. (Eds.), Vertebrates
in the tropics, Museum Alexander Koenig, Bonn,
pp. 321–332.
Hepburn H.R., Radloff S.E. (1998) Honeybees of
Africa, Springer-Verlag, Berlin.
21
H.R. Hepburn et al.
Meixner M., Sheppard W.S., Dietz A., Krell R. (1994)
Morphological and allozyme variability in honey
bees from Kenya, Apidologie 25, 188–202.
Muzaffar N., Ahmad R. (1989) Distribution and com-
petition of Apis spp. in Pakistan, Proc. 4th Int. Conf.
Apic. Trop. Clim., Cairo, pp. 449–452.
Narayanan E.S., Sharma P.L., Phadke K.G. (1960)
Studies on biometry of the Indian bees, 1. Tongue
length and number of hooks on the hind wings of
Apis indica F., Indian Bee J. 22, 58–63.
Narayanan E.S., Sharma P.L., Phadke K.G. (1961a)
Studies on biometry of the Indian bees, III. Tongue
length and number of hooks on the hind wings of
Apis indica F. collected from Madras State, Indian
Bee J. 23, 3–9.
Narayanan E.S., Sharma P.L., Phadke K.G. (1961b)
Studies on biometry of the Indian bees, IV. Tongue
length and number of hooks on the hind wings of
Apis indica F. collected from Uttar Pradesh, Indian
Bee J. 23, 69–74.
Nunamaker R.A., Wilson W.T., Ahmad R. (1984)
Malate dehydrogenase and non-specific esterase
isozymes of Apis florea, A. dorsata and A. cerana
as detected by isoelectric focusing, J. Kans. Ento-
mol. Soc. 57, 591–595.
Ono M. (1992) The Asian honeybees (Apis spp.), Hon-
eybee Sci. 13, 19–22 (in Japanese).
Otis G.W., Hadisoesilo S. (1990) Honeybee survey of
South Sulawesi, J. Penelitian Kehutanan 4, 1–3.
Otis G.W. (1991) A review of the diversity of species
within Apis, in: Smith D. (Ed.), Diversity in the
Genus Apis, Westview, Boulder, pp. 29–49.
Peng Y.S., Nasr M.E., Locke S.J. (1989) Geographical
races of Apis cerana Fabricius in China and their
distribution. Review of recent Chinese publications
and a preliminary statistical analysis, Apidologie
20, 9–20.
Pesenko Y.A., Lelej A.S., Radchenko V.G., Filatkin
G.N. (1990) Chinese wax-bee Apis cerana cerana
(Hymenoptera, Apidae) in the Soviet Far East, Ento-
mol. Rev. 69, 21–46.
Radloff S.E., Hepburn H.R. (1998) The matter of sam-
pling distance and confidence levels in the sub-
specific classification of honeybees, Apis mellif-
era L., Apidologie 29, 491–501.
Rinderer T.E., Koeniger N., Tingek S., Mardan M.,
Koeniger G. (1989) A morphological comparison of
the cavity dwelling honeybees of Borneo Apis
koschevnikovi (Buttel-Reepen, 1906) and Apis cer-
ana (Fabricius, 1793), Apidologie 20, 405–411.
Rozalski R.J., Sakurai H., Tsuchida K. (1996) Esterase
and malate dehybrogenase isozymes analysis in the
population of honeybee, Apis cerana japonica and
Apis mellifera, Jap. J. Entomol. 64, 910–917.
Ruttner F. (1988) Biogeography and Taxonomy of
Honeybees, Springer-Verlag, Berlin.
Ruttner F. (1992) Naturgeschichte der Honigbienen,
Ehrenwirth, München.
Ruttner F., Tassencourt L., Louveaux J. (1978) Bio-
metrical statistical analysis of geographic variabil-
ity of Apis mellifera L., Apidologie 9, 363–381.
Ruttner F., Kauhausen D., Koeniger N. (1989) Posi-
tion of the red honey bee, Apis koschevnikovi
(Buttel-Reepen 1906), within the genus Apis,
Apidologie 20, 395–404.
Ruttner F., Pour Elmi M., Fuchs S. (2000) Ecoclines in
the Near East along 36° N latitude in Apis mellifera
L., Apidologie 31, 157–165.
Sakai T. (1956) Morphological studies on the difference
among some strains of the honeybees, Honeybee
Sci. 9, 115–125 (in Japanese).
Sakai T. (1958) Morphological studies on the drone
honeybees, Jap. Bee J. 11, 40–46 (in Japanese).
Sasaki M. (1994) Comparative evaluation of southern
and northern Apis cerana ecotypes and points which
should be considered for future breeding, Honeybee
Sci. 15, 99–106 (in Japanese).
Schneider P., Djalal A.S. (1970) Vorkommen und Hal-
tung der Östlichen Honigbiene (Apis cerana Fabr.)
in Afghanistan, Apidologie 1, 329–341.
Schneider P., Kloft W. (1971) Beobachtungen zum
Gruppenverteidigungsverhalten der Östlichen
Honigbiene Apis cerana Fabr., Zool. Tier. 29,
337–342.
Sheppard W.S., Berlocher S.H. (1989) Allozyme vari-
ation and differentiation among four Apis species,
Apidologie 20, 419–431.
Sihanuntavong D., Sittipraneed S., Klinbunga S. (1999)
Mitochondrial DNA diversity and population struc-
ture of the honey bee, Apis cerana, in Thailand,
J. Apic. Res. 38, 211–219.
Singh M.P., Verma L.R., Daly H.V. (1990) Morpho-
metric analysis of the Indian honeybee in the north-
east Himalayan Region, J. Apic. Res. 29, 3–14.
Smith D.R. (1991) Mitochondrial DNA and honey bee
biogeography, in: Smith D. (Ed.), Diversity in the
Genus Apis, Westview, Boulder, pp. 131–176.
Smith D.R., Glenn T.C. (1994) Allozyme polymor-
phisms in Spanish honey bees (Apis mellifera iber-
ica), J. Hered. 86, 12–16.
Smith D.R., Hagen R.H. (1997) The biogeography of
Apis cerana as revealed by mitochondrial DNA
sequence data, J. Kans. Entomol. Soc. 69, 294–310.
Smith D.R., Hagen R.H., Otis G.W. (1999) Phylogeny
and biogeography of Apis cerana subspecies: test-
ing alternative hypotheses, in: Hoopingarner R.,
Connor L. (Eds.), Apiculture for the 21st Century,
Wicwas Press, Cheshire, CT, pp. 60–68.
Smith D.R., Palopoli M.F., Taylor.B.R., Garnery L.,
Cornuet J.M., Solignac M., Brown W.M. (1991)
Geographical overlap of two mitochondrial
genomes in Spanish honey bees (Apis mellifera
iberica), J. Hered. 82, 96–100.
Smith D.R., Villafuerte L., Otis G., Palmer M.R (2000),
Biogeography of Apis cerana F. and A. nigrocincta
Smith: Insights from mtDNA studies, Apidologie
31, 265–280.
22
Infraspecific categories of Apis cerana
Verma L.R., Kalle G.P., Sharma A., Mattu V.K. (1989)
Biometry of Apis cerana of Nepal, Himalayas, Proc.
4th Int. Conf. Apic. Trop. Clim., Cairo, pp. 458–465.
Verma L.R., Mattu V.K., Daly H.V. (1994) Multi-
variate morphometrics of the Indian honeybee in
the northwest Himalayan region, Apidologie 25,
203–223.
Whitmore T.C. (1984) Tropical rain forests of the Far
East, second edition, Oxford Scientific Publica-
tions, Clarendon Press, Oxford.
Yang G.H., The resources of the Chinese honeybee
and its utilization (cited from Peng et al., 1989).
Zhen-Ming J., Guanhuang Y., Shuangxiu H., Shikui
L., Zaijin R. (1992) The advancement of apicul-
tural science and technology in China, in: Verma L.
(Ed.), Honeybees in Mountain Agriculture, Oxford
& IBH, New Delhi, pp. 133–147.
Sylvester H.A., Limbipichai K., Wongsiri S., Rinderer
T.E., Mardan M. (1998) Morphometric studies of
Apis cerana in Thailand and the Malaysian penin-
sula, J. Apic. Res. 37, 137–145.
Tanabe Y., Tamaki Y. (1985) Biochemical genetic
studies on Apis mellifera and Apis cerana, Proc.
30th Int. Beekeep. Congr., Nagoya, Japan,
pp. 152–154.
Tilde A.C., Fuchs S., Koeniger N., Cervancia C.R.
(2000) Morphometric diversity of Apis cerana Fabr.
within the Philippines, Apidologie 31, 249–264.
Tokuda Y. (1924) Studies on the honey bee, with spe-
cial reference to the Japanese honey bee, Trans.
Sapporo Nat. Hist. Soc. 9, 1–26.
Verma L.R. (1990) Beekeeping in Integrated Moun-
tain Development, Oxford & IBH, New Delhi.
Verma L.R. (Ed.) (1992) Honeybees in Mountain Agri-
culture, Oxford & IBH, New Delhi.
23
To access this journal online:
www.edpsciences.org
