Apidologie 31 (2000) 313–339 313 © INRA/DIB-AGIB/EDP Sciences

Review article

Reproductive isolation among of the Apis

Nikolaus KOENIGER*, Gudrun KOENIGER

Institut für Bienenkunde, (Polytechnische Gesellschaft), Fachbereich Biologie der J.-W. Goethe Universität Frankfurt am Main, Karl-von-Frisch Weg 2, 61440 Oberursel, Germany

(Invited paper)

Abstract – In the 1960s, research on reproductive isolation in honeybees started with the pioneering work on and A. mellifera of F. Ruttner. Since then, the number of recognised Apis species increased from four to nine, and data on reproductive isolation played a key role in this development. In this paper, we discuss the behavioural mating barriers (mating season, mating place, sexual signals, daily mating periods), copulatory barriers (size, genitalia, mating sign) and physiological barriers (sperm transfer, sperm storage) and postzygotic barriers (fertilisation, development, hybrids). Allopatric A. mellifera and allopatric populations of the other species had a uniform mating period dur- ing the afternoon hours. Sympatric honeybee species were separated mainly by different daily mat- ing periods. The mating period differed between populations of the same species from different regions. The sequence of the mating periods, however, described from , and () followed the same pattern and showed a taxonomic and size correlation: the dwarf bees (A. andreniformis and/or ) occupied the first position shortly after noon. The next mat- ing period was occupied by cavity-dwelling bees and at sunset, A. dorsata drones flew out for mat- ing. In addition, in the honeybee species that have been studied, various non behavioural mating barriers have been demonstrated. reproductive isolation / Apis / mating behaviour / genitalia / hybrid

1. INTRODUCTION systematics [8, 23, 50]. In spite of these dif- ferences, however, traditional definitions Among evolutionary biologists, defini- acknowledge the central importance of tions of ‘what a species is’ are highly reproductive isolation. Effective barriers to diverse. Furthermore, different species con- between populations prevent any cepts have had a considerable influence on possibility of subsequent reintegration. Thus

* Correspondence and reprints E-mail: [email protected] This work is dedicated to the memory of the late Friedl Ruttner, whose unfailing enthusiasm for research on honeybee reproduction has been an inspiration to us. 314 N. Koeniger, G. Koeniger achieving reproductive isolation becomes barriers operate early, and prevent the phys- an evolutionary ‘key event’ which marks a ical contact between queen and ; point of independent genetic development ii) copulatory barriers operate during the and divergence. process of copulation (from the first con- In honeybees, research into reproductive tact to the separation of the queen and the isolation has had a relatively recent start. drone). These mechanisms prevent sperm The Western honeybee species Apis melli- transfer to the queen; iii) physiological bar- fera L. with its wide, natural distribution in riers operate after copulation. These barriers Europe, in the Middle East and Africa, and block the path of the sperm from the queen’s its successful introduction into the Americas oviduct into the spermatheca and further to and Australia has dominated scientific inter- the fertilisation of the egg; iv) postzygotic est. Since the start of modern barriers disturb normal development, result- in Europe [10] about 150 years ago, numer- ing in the death or in the infertility of hybrids. ous honeybee colonies and queens have been imported from various places (Africa, the Middle East, the Mediterranean islands). 2. BEHAVIOURAL BARRIERS These exotic bees, crossed with the local population, resulted in multiple hybrids, all 2.1. Reproductive isolation due to of them fertile between one another without seasonally different mating periods any apparent limitation [74]. This ‘absence’ of reproductive isolation lasted until 1965, Monogyny is one of the basic features of when A. cerana Fabricius colonies from the honeybee colony. This links the rearing were imported into Germany, and the of new queens to the process of colony mul- first pioneering research on reproductive tiplication [69]. In other words, the mating isolation among honeybees was initiated by season in honeybees is inevitably linked to F. Ruttner. the season. Reproductive swarm- Traditionally, isolation mechanisms are ing depends on favourable environmental categorised according to their temporal rela- conditions. Specifically, ample and tion to fertilisation as prezygotic − or postzy- must be available for two reasons: to gotic barriers [50]. To accommodate the produce enough bees before colony fission, special mating biology of honeybees and and to support the swarms which do not the complicated process of multiple mating have combs or any storage at the in the genus Apis [27], we divided the beginning [84]. For survival, a new swarm prezygotic phase into three subgroups of needs more or less immediate access to nec- isolating factors. The following definitions tar and pollen for comb building and brood apply to this paper (Tab. I): i) behavioural rearing. Otherwise the natural mortality of

Table I. Categories of mating barriers and ‘factors’ which may cause reproductive isolation in the genus Apis.

Behavioural Copulatory Physiological Postzygotic barriers barriers barriers barriers (prezygotic) (prezygotic) (prezygotic)

Mating season Size Sperm transfer Fertilisation Mating places Genitalia Sperm storage Hybrids Sexual signals Mating sign Daily mating periods Reproductive isolation among species of the genus Apis 315 workers cannot be compensated and later During mating flights, A, mellifera drones the swarm (new colony) is reduced to congregate in the open air above their drone beyond the critical threshold. Therefore, the congregation area [106] where they remain, mating season in honeybee populations flying in wide loops until they return to the depends on seasonal blooming cycles. This colony to feed [75]. Congregation areas usu- holds true for allopatric A. mellifera in ally have a diameter of 30–200 m. More Africa and Europe [74], and for populations recently, the area above which the drones of sympatric Asian species. Accordingly, flew was measured by radar as 1 600 m2 in Sri Lanka [39], in Thailand [67] and in [48]. Nonetheless, a congregation area has a Borneo [44] all sympatric Apis species pro- limited spatial extension and A. mellifera duced drones simultaneously. We assume, drones are not attracted by a queen flying because of the uniform mode of colony mul- outside the area [75]. Depending on weather tiplication by swarming within the genus, conditions, A. mellifera drone populations that there is not much ‘evolutionary flexi- fly at a preferential height above the ground bility’ to change the reproductive season that varies from 5–40 m above the ground. between sympatric honeybee species. Within a single drone congregation area, A. mellifera ligustica drones and A. mellifera carnica drones showed a distinct difference 2.2. Reproductive isolation due in vertical distribution: A. m. ligustica drones to different mating places were more frequently caught at lower alti- tudes (4 m above ground), while A. m. car- Among the Apoidea we find an impres- nica drones preferred to fly at higher alti- sive variability of locations where mating tudes [29]. Virgin queens that mated with occurs. Several solitary species mate at the that mixed drone population subsequently nesting sites. Often the males emerge earlier produced significantly more ‘pure’ worker and assemble (and compete) at the nesting offspring: Thus, A. m. ligustica queens sites where they mate with freshly emerging mated more often with A. m. ligustica drones females. In the case of species-specific nest- and A. m. carnica queens mated preferably ing sites, such behaviour would serve as a with A. m. carnica drones. The above dif- mating barrier. Other Apoidea mate on flow- ferences between cruising altitudes of A. m. ers which, in the case of oligolectic bees, ligustica and A. m. carnica drones were dis- may lead to the ‘rendezvous’ of conspecific cussed as a mechanism of assortative mating mates [1]. Another example of spatial dif- (i.e., incomplete reproductive isolation) [29]. ferentiation of mating places comes from Assortative mating has been reported to sympatric Bombus species. The males build occur in a mixed population of A. m. ligus- an odour track by deposition of pheromones tica and A. m. scutellata in South America which attract virgin queens. The altitude [25]. (height) of these odour tracks is different, Since virgin queens commence their mat- ranging from the ground to the canopy [20]. ing flights significantly later than drones, Thus spatial separation seems to work in the congregation of drones is formed irre- favour of reproductive isolation. spective of the presence of a queen [29, 79]. In discussing the genus Apis, we shall The same drone congregation area was vis- start with a short description and review of ited by A. mellifera drones each season for the drone congregation areas of honeybee more than 30 years ([29]; Pechhacker, species. This is intended to lay the basis for unpubl. data). Near Selbourne (England) a the question of whether or not a spatial sep- drone congregation area has been recorded aration of mating locations can function as for 197 years [95]. In A. mellifera, several a behavioural barrier between honeybee drone congregations have been found within populations. the flight range of a colony [77, 78, 106]. 316 N. Koeniger, G. Koeniger

Drones of A. mellifera were found visiting a tree canopy might have a higher selective drone congregation area at a distance of up advantage under tropical conditions than in to 8 km from their colonies, and drones from the mountains of Northern [63]. a considerable number of different colonies Considering the limited data available (in and apiaries were found at drone congrega- comparison to A. mellifera), we expect that tion areas [75]. Calculations of the related- an even wider range of differences among ness of drones captured in a congregation the drone congregation areas of A. cerana area in Germany revealed that these drones may become apparent with further research originated from about 240 different colonies, on Asian honeybees. and A. mellifera probably represents one of Drone congregations of A. koschevnikovi the most elaborate panmictic systems pos- were regularly observed to occur under a sible among terrestrial [3]. The thick cover of vegetation, and the height physiographical structure of A. mellifera above the ground of different drone con- drone congregation areas seems to vary gregation areas varied between 1.5 to 12 m greatly [77]. In plains and less structured [46]. At present, there is no information areas, however, drones have been reported available on the drone congregation areas to be distributed more uniformly, and were of the other two cavity-dwelling honeybees, attracted to queens wherever they were A. nuluensis and A. nigrocincta. placed [4, 47, 95]. The local conditions that make the A. mellifera drones stay or return In Borneo, drones of A. dorsata congre- to a drone congregation area remain gate under the canopy of tall emergent trees. unknown [43, 83]. The eminent, tall tree tops seem to serve as a visual landmark, and applying this crite- In contrast to A. mellifera drones which rion, several ‘new’ A. dorsata drone con- congregate in the open air, A. cerana indica gregation areas have been located [43]. drones in Sri Lanka and in Borneo gather Recently, several A, dorsata drone congre- in close proximity to the trees. These drones gation areas have been detected under tall restrict their flight to an open space within or trees in Sri Lanka (Punchihewa, unpubl. near the canopies of the trees. They do not data). The drones of A. dorsata assemble follow a (caged) queen far into the open under the umbrella of the canopy, and do space above or at the side of the canopy [46, not follow the queens which move into the 64]. The distance between the drone con- open air. Further, drone attraction showed a gregation area and the drone colonies is maximum of 3–5 m below the canopy. The clearly less than in A. mellifera and ranges height above the ground ranged between 10 up to 2 km (Tingek, unpubl. data). In Japan, and 35 m depending on the size of the tree however, drones of A. c. japonica congre- [43]. The drone congregation area of the gate in the open air high above prominent other giant honeybee species, A. laboriosa, trees [13, 104]. In Germany, A. c. indica remains undiscovered. Also the drone con- drones originating from Northern Pakistan gregation areas of the dwarf honeybees, visited a drone congregation area in an open A. andreniformis and A. florea, have not yet valley far away from the trees [71, 80]. been found. Taken together, drone congregation areas of A. cerana show a high degree of variability. It is not surprising that among allopatric Since these differences can occur within the species, drone congregation areas show sim- same (A. c. indica), the specific ilarities. Some convincing evidence for these features of these drone congregation areas similarities came from Ruttner [71]. In Ger- may be mainly the result of local adapta- many, drones from imported A. c. indica tions to environmental factors. For exam- colonies (which originated from Pakistan) ple, avoiding predators such as were caught together with simultaneously (Merops sp., etc.) by flying near or within a flying A. m. carnica at the same drone Reproductive isolation among species of the genus Apis 317 congregation area. An estimation of drone ing our 10-year study period, we located numbers in the nearby colonies (A. cerana 5–10 colonies of this species each season. and A. mellifera) showed that the ratio of Because of the small size and the hidden A. cerana drones at that drone congregation nest sites, we assume that this was only a area was similar to that of A. mellifera rather small proportion of a sizeable popu- drones. A second case of heterospecific lation of A. andreniformis in the area. How- drone mixing, although to a lesser extent, ever, drone congregation areas of A. andre- was recorded from drone congregation areas niformis were not detected. in Japan. In that study, a small number of heterospecific drones was caught at each A general scheme is presented (Fig. 1), congregation area [104]. which is mainly based on observations from At present, the only comparative obser- three different drone congregation areas vations and experiments on drone congre- [46]. At these places the major landmark gation areas of sympatric Asian honeybees was an outstanding tree top which clearly have been carried out in Borneo (Tenom). protruded from the horizon line. Drones of So we must restrict the following section to A. c. indica had their maximal flight fre- the situation in Tenom. Further, we caution quency measured by attraction to our stan- that the conclusions remain more or less dard dummies (Fig. 2) slightly outside the preliminary until confirmed by investiga- canopy of the trees and larger shrubs, about tions from other places. During our experi- 10 to 12 m above the ground. When dis- ments, fairly large numbers of honeybee turbed by birds or by the net, A. cer- colonies of four species were found. A. cer- ana drones escaped rapidly into the cover ana indica and A. koschevnikovi colonies of the branches. Drones of A. koschevnikovi were kept in modern hives, and both species remained under the dense cover of the supplemented a larger natural population of canopy, and flew in a space 6 to 8 m above feral colonies. About 50 to 100 A. dorsata the ground. A. dorsata drones flew under colonies were found nesting in several bee the first layer of branches at a height of 20 to trees near (within a radius of 5 km) the exper- 25 m. So the distribution of drones resulted imental area. A. andreniformis foragers were in a clear spatial separation without any frequently observed on various flowers. Dur- overlap between the three species.

Figure 1. Drone congre- gation areas in Sabah (Bor- neo). A. dorsata drones congregated directly under the canopy of high emer- gent trees. A. koschevnikovi drones congregated under the thick cover of branches and trees. A. cerana drones remained near the branches of neighbouring vegetation. All three drone congrega- tion areas were within a distance of 30 m. Spaces occupied by drone flight. 318 N. Koeniger, G. Koeniger

Figure 3. Copulation dummy with two Figure 2. An A. koschevnikovi drone attracted A. koschevnikovi drones. At both ends of the to the standard dummy. A section of black pen- standard dummy (1 mg 9-ODA) the hollowed cil (length 34 mm, diameter 8 mm) was tied to a adomen of an Apis queen was glued. The thread and impregnated with 1 mg 9-ODA. A. koschevnikovi drones were attracted, copu- lated and became stuck in the abdomen.

But was this separation sufficient to which was impregnated with 1 mg 9-ODA. ensure a complete reproductive isolation This quantity is about 5 to10 times higher among the species? Initially, we carried out than the amount of pheromone of a young some unintentional experiments: Lifting the A. mellifera queen [86]. In two cases, how- standard dummy (Fig. 2) to a height of 25 m, ever, we had the opportunity of comparing we observed a few A. koschevnikovi drones the dummy to a natural free-flying A. cerana pursuing the dummy at a height of 1.5 to queen: the queen was found to be far more 20 m above the ground, and single drones attractive − the drones left the dummy and of A. koschevnikovi were attracted even to chased the queen. a height of 25 m (optimal for A. dorsata So, the above observations indicate that drones). Later, we used copulation dummies the spatial separation between drone (Fig. 3) with an opening from which the congregation areas does not function as a copulating drone could not extricate him- complete reproductive isolation mechanism self, and was pulled down for careful exam- between sympatric honeybee species in ination [43]. A few A. koschevnikovi drones Borneo. were caught at the optimum of the A. cerana drone congregation area and at the A. dor- sata drone congregation area. A. cerana 2.3. Different sexual signals as means drones were caught at the A. dorsata drone of reproductive isolation congregation area, but not at the A. koschev- nikovi drone congregation area. Apis dor- The highly developed social life of hon- sata drones came to the drone congregation eybees has affected mating behaviour in area of A. cerana, but not to the densely several ways. The sex ratio is strongly male- covered drone congregation area of A. kos- biased; colonies produce some hundred chevnikovi. As a result of these experiments, times more drones than queens [69]. Tak- single drones were attracted and copulated ing into account the fact that Apis drones at a drone congregation area of another perform multiple nuptial flights while a species (with the exception of the A. koschev- queen flies only once or twice for mating, nikovi drone congregation area). the effective male bias at a drone congre- A critical argument in this discussion gation area is further increased. As a con- must address the standard dummy (Fig. 2), sequence, male-male competition [93] must Reproductive isolation among species of the genus Apis 319 have a major impact on drone behaviour. In impregnated with 1 mg of 9-ODA success- pursuit of a flying virgin queen, the speed fully attracted drones of A. cerana [64, 71, (time) for a drone to reach his target has the 104] A. dorsata [43] and A. koschevnikovi highest priority. Under these conditions, [46]. Underlining the uniform mechanism drones depend on simple and rapidly of drone attraction further, Gries [18] detectable stimuli for queen recognition. attracted drones of A. dorsata, A. mellifera, Further, in order to obtain their goal, an A. cerana and A. koschevnikovi to the same immediate reaction to a first, albeit uncertain dummy. Drones were not only attracted but signal of a virgin queen is the better choice, also started to grasp the dummy and to ini- because the probability of a drone encoun- tiate copulation. However, for successful tering a second queen is close to zero [19]. copulation an opening at the end of the The first reaction of drones to visual stim- dummy (like an open sting chamber) is uli seems to depend on ‘unspecific’ move- required (Fig. 3) [16]. ments. At a congregation area, drones of Plettner et al. [62] found specific differ- A. mellifera react to various moving objects ences in the mandibular gland signals by a fast, short turning reaction. Flying birds, between queens of A. mellifera, A. dorsata, butterflies and even stones thrown into the air A. florea and A. andreniformis. The ques- area will momentarily attract some drones tion must be addressed whether or not the [24]. Without the presence of pheromones, results of the behavioural experiments however, these objects will never initiate any pursuit or consistent attraction. Accord- explore the natural situation. We cannot ing to Strang [90], the optimal surface of a exclude the fact that the demonstrated inter- queen dummy (impregnated with phero- specific drone attraction was caused by our mone: 5 mg 9-ODA) was slightly larger than experimental techniques: for example, that of a queen (about 3 cm2). With further excessive use of 9-ODA or any other super increase in size, the attraction to drones stimuli of the dummies. decreased. Black (and red) was the optimal Direct observations and tests of inter- colour. In other words, the visual signal specific reactions were carried out with attracting A. mellifera drones seems to be A. mellifera and imported A. cerana in Ger- rather unspecific: fast moving, dark objects many by Ruttner and Kaissling [76]: elec- of a size slightly larger than a queen seem to trophysiological responses of receptors (sen- be nearly optimal. silla placodea) on the antenna of the The major active chemical component A. cerana and A. mellifera drone showed of the A. mellifera queen’s mandibular gland no differences between extracts of mandibu- was identified as (E)-9-oxo-2-decenoic acid lar glands of A. cerana and A. mellifera (9-ODA) by Callow and Johnston [6]. queens. However, at a drone congregation Among several other important biological area in Germany, significantly more A. mel- functions, 9-ODA was found to be the main lifera drones were attracted to conspecific component of the A. mellifera queen’s sex (caged) queens than to caged A. cerana attractant [15, 59]. Later, it was demon- queens. These results indicate that the strated that extracts of queens of three other species-specific differences in queen Apis species (A. florea, A. cerana and pheromones between A. mellifera and A. dorsata) attract A. mellifera drones, and A. cerana (as described by Plettner et al. that these extracts contained 9-ODA. A. dor- [62]) might be of functional significance. sata and A. cerana queens had a quantity However, the preference for conspecific (150 to 300 µg) of 9-ODA similar to that queens does not prevent the mating of a het- of the A. mellifera queen [5, 87]. erospecific queen. Ruttner and Maul [79] Consequently, dead, extracted queens or reported a young A. cerana queen which black dummies of similar size (Fig. 2) had her bursa copulatrix blocked by a 320 N. Koeniger, G. Koeniger mating sign of an A. mellifera drone (easily A. mellifera queens perform their mat- recognised by its chitinous plate). This is – ing flights during the peak period of drone as far as we know – the only direct evidence flight. Successful mating flights of queens of heterospecific mating of a free-flying (returning with a mating sign!) occurred in Apis queen! Austria between 14.20 and 16.10 [29]. In Africa, A. m. monticola mated queens Apparently, the drastic male bias at the returned between 13.00 and 15.30 [47]. drone congregation area and the resulting competition among drones has led to a fast, Overall, the mating flight period of A. mel- simple and uniform mechanism of queen lifera invariably starts shortly after noon recognition. The main olfactory signal and and covers a period of 4 to 5 h. The differ- essential sex pheromone seems to be 9-ODA ences reported so far have been within a in all Apis species. The specific differences short range and do not exceed ± 1 hour. The in the queen’s pheromone spectrum among period of actual mating (as documented by the species [62] may result in a reduced queens returning with a mating sign) is con- attraction of the heterospecific queen. It will, siderably shorter (2 to 3 h). Queens fly later however, not prevent interspecific copula- and stop their flight activity earlier than tions as Ruttner and Maul [81] demon- drones. Compared to other behavioural char- strated. This leads to the conclusion that the acters (defence, swarming etc.), the daily differences in sexual signals do not play a mating period of A. mellifera seems to be major role as a behavioural barrier between rather uniform in Africa and Europe. sympatric honeybee species. 2.4.2. Apis cerana: the ‘allopatric situation’ 2.4. Reproductive isolation due to different daily mating periods Among the Asian honeybee species, A. cerana has the most extensive natural 2.4.1. Apis mellifera: distribution. In consequence, it overlaps the ‘allopatric situation’ many of the other Asian Apis species [73]. Regionally, there are, however, large areas As discussed before, the natural distri- where A. cerana is the only honeybee. Within the Asian continent these areas are bution of A. mellifera generally does not mainly in the northern part of their range, overlap with the distribution of other Apis in mountain ranges and in the Japanese species [73]. Thus for A. mellifera, we can islands (with the exception of Hokkaido). assume that an evolution “under the condi- tion of being the only honeybee species at a A daily A. cerana mating period from place” has shaped the mating behaviour and 12.30 to 16.00 has been reported from Bihar the daily mating period. Drones of Euro- in North [85]. Drones of A. c. indica pean races of A. mellifera start flying shortly (originating from the mountains of the North after the sun passes the zenith (12.15), and West Frontier Province of Pakistan) flew stop in the late afternoon (17.00) [70]. In between 12.00 and 15.30 in Germany [81]. Germany, a comparison of drone flight times Verma [98] observed mating flights of A. between A. m. ligustica and A. mellifera car- c. indica queens in the Shimla Hills (North nica showed no significant differences [9, India) between 12.30 and 15.30. In Japan, 29]. Observations in Africa, near Pretoria, drones of A. c. japonica flew from 13.15 to with A. m. scutellata showed a period from 17.00, and successful mating flights of 12.45 to 16.45 [95], and more recently A. c. japonica queens occurred between Lahner [47] reported A. m. monticola drone 14.35 and 16.35 [105]. flight activity in Malawi to occur from 11.20 The mating period of A. mellifera and to 16.00. the observations from regions where Reproductive isolation among species of the genus Apis 321

A. cerana occurs as the only Apis species The drone flight period of A. c. indica drones shows a striking degree of similarity. The (Tab. II) in Sri Lanka was confirmed by overall duration of drones’ and queens’ Punchihewa et al. [64]. Accordingly, queens flights and the timing during the early successfully mated between 16.15 and 16.55 afternoon seems to be nearly identical in [63, 65]. The mating period of A. c. indica in A. mellifera and in ‘allopatric’ populations Sri Lanka is the latest so far recorded for of A. cerana. this species.

2.4.3. The ‘sympatric situation’ A. koschevnikovi drones fly during a long period of nearly 2 hours. Queens flew The following table (Tab. II) focusses on between 17.00 and 18.15 [33]. A. dorsata observations and research which present fly consistently at sunset. The flight period data on mating flights of sympatric species of A. dorsata drones is very short. Drones of at one location. this species perform daily only a single flight A. andreniformis drones (Tab. II) had a [43]. In Borneo, a slight overlap with uniform short flight period after 12.00. The A. koschevnikovi drones occurred. This was, period of A. andreniformis queen flights however, too slight to affect the reproductive was between 12.33 and 12.50 in Sabah, isolation. Borneo [38]. Koeniger and Wijayagunasekera [39] A. florea drones (Tab. II) show remark- observed that Apis florea in Sri Lanka occu- able differences. In Sri Lanka, drones flew pies the time window which is nearest to earlier than in South East Thailand. Accord- the ‘allopatric’ mating period. Therefore, ing to Koeniger et al. [44], the drone flight they argued that A. florea was the original period of A. florea in Bangkok was between Apis species to arrive at Sri Lanka. Conse- 13.45 and 15.30, and ended more than quently, the species with later periods would 1 hour earlier compared to the following have established colonies in Sri Lanka some data (Tab. II). Also, successful mating flights time later. With the evidence (Tab. II) avail- of A. florea queens in Bangkok were able today, however, a general and uniform observed between 14.04 and 14.25 [42]. pattern related to becomes appar- Apparently, some variability of A. florea ent: the first position directly after noon is mating periods occurs in Thailand. Perhaps held by a dwarf bee species (A. andreni- the earlier mating period (in comparison to formis and/or A. florea). The next time win- Tab. II) is typical for regions (like Bangkok) dow seems to belong to one or even two where A. florea is the only dwarf bee species. cavity-dwelling species (A. cerana and A. cerana exhibits more variability in A. koschevnikovi); and at the very end of drone flight period than other Apis species. the day, just around sunset, A. dorsata holds

Table II. Drone flight periods of sympatric Asian honeybee species.

1st author (year) Koeniger [39] Rinderer [67] Koeniger [45] Locality Sri Lanka Thailand Sabah, Borneo

A. andreniformis 12.15–13.45 12.00–13.45 A. florea 12.00–14.30 14.00–16.45 A. cerana 16.15–17.15 15.15–17.30 14.00–16.15 A. koschevnikovi 16.45–18.30 A. dorsata 18.00–18.45 18.15–18.45 18.15–19.05 322 N. Koeniger, G. Koeniger its mating time. It seems to be unlikely that habitats. However, in the case where they the similarity in this sequence of mating spread into a ‘new’ territory alone, each periods originated by chance in three dif- species will shift towards the ‘allopatric’ ferent locations (Sri Lanka, Thailand and mating period. Several ‘mountain’ bee pop- Sabah). Another possibility, that this ulations may serve as an example for the sequence evolved once in South East latter phenomenon: the allopatric popula- and spread unchanged to the above places, tion of A. cerana from Northern India and seems to be equally unlikely. Therefore we Pakistan has already been discussed above. hypothesise that this pattern originated from Further, the open-nesting giant honeybee a (yet unknown) mechanism of interspecific species A. laboriosa, adapted to the high reproductive competition, which causes pre- altitudes of the Himalayas, has a drone flight dictable results independently of place and period between 12.20 and 14.20 [97]. Like- environment. Without exception, the tem- wise, A. nuluensis is the only honeybee poral sequence of mating periods is strictly species in the mountains of Borneo above correlated with the size; it starts with the 1 700 m, and its drone flight period is smallest Apis species (A. andreniformis) between 10.45 and 13.15 [44]. No observa- and ends with the largest sympatric species tions on queen flights are yet available, and (A. dorsata)! as the drone flight was recorded from one colony only (this applies to A. laboriosa It is rather tempting to speculate on how too), there is need of additional confirma- drone behaviour can result in this sequence: tion. The basic reaction of Apis drones is directed to queens which are larger than fellow Arguably, the time-sharing mating pat- drones. So smaller drones may try to tern of sympatric Apis species has evolved to copulate unidirectionally with larger drones, form a nearly perfect behavioural barrier. excluding them from access to queens. How- Regarding the temporal pattern of Apis mat- ever, these speculations are premature, and ing behaviour, a separate daily mating period the question of whether the size had a direct becomes operational earlier than several effect, or rather size-correlated factors complicated and ‘risky’ events in a behav- caused the above sequence, remains ioural sequence. After this successful ‘a pri- unsolved. ori’ reproductive isolation has been estab- lished, previously functional mechanisms Our above hypothesis, however, has which operate on later steps in mating come to a large-scale test. Recently, A. flo- behaviour become less meaningful. In other rea was involuntarily introduced into Africa words, flying during their daily mating [53]. Fairly large populations established in period, drones and queens of any Apis the region of Khartoum (where no feral species do not gain further by maintaining A. mellifera exist) and A. florea colonies are different sexual signals or species-specific spreading to the South, along the Nile [52]. mating places. Eventually A. florea will reach the habitat of A. m. scutellata. In the likely event of a What is involved in the timing of mating sympatric co-existence, we predict that as flights? Taber [91] confined an A. mellifera colony in a cool dark room for 12 hours, a result of a rapid natural selection process inducing earlier drone flight on the follow- A. florea drones will fly prior to A. mellifera ing day. Yoshida and Yamazaki [104] were drones! able to shift the flight period of A. mellif- Whenever the sequence of separated mat- era drones by changing the photoperiod. ing periods was established, it facilitated The above results suggest that the drone sympatric co-existence, and Apis species flight period seems to depend on an internal could spread simultaneously, sharing their clock. Koeniger et al. [34] used cross-foster Reproductive isolation among species of the genus Apis 323 techniques. They introduced drones of 3.1. Reproductive isolation A. koschevnikovi and A. cerana into alien due to different size colonies. As a result, A. koschevnikovi drones flew during their species-specific A well-known example for size differ- mating period independently of their A. cer- ences as a copulatory barrier comes from ana host colony. Similarly, A. cerana drones domestic animals. For example, females and followed their own mating period and not males of different dog breeds are attracted to the A. koschevnikovi colony’s timetable. each other, but because of differences in Later, virgin A. koschevnikovi queens were size are unable to copulate. introduced into A. cerana colonies and flew In Apis species, worker bees vary con- at their own species’ mating period [44]. siderably in size [84]. The weight relation of Drones and queens seem to decide on the workers of the dwarf honeybee species and right time for mating on the basis of an the Asian cavity-dwelling species is about ‘inherited timetable’. Consequently, a direct 1 to 2.5; to the giant honeybees it is about evolutionary impact on the individual drone 1 to 5 [32]. But the difference in weight in or queen and selection for changes in mating drones in these species is much smaller. The period becomes operational and may act relation between the dwarf and the Asian faster than any effect via the colony (work- hive-dwelling bees increases gradually to ers). Thus, fast adaptations to predatory 1.5, and compared to the giant honeybees pressure, to other environmental alterations it is about 1 to 2 (Tab. III). Only the drone of or even to ‘new’ honeybee species are facil- the allopatric A. mellifera has a consider- itated. ably higher weight. The weights of queens have about the same relation as in drones. As previously mentioned, drones of all 3. COPULATORY BARRIERS species will grasp the same queen dummy and try to copulate. Considering the small Many cases are known in which different size differences (especially within the same sizes and shapes of genitalia do not totally taxonomic group), we conclude that body prevent copulation. But then interspecific size ‘per se’ does not play a major role as copulation often results in injury or even an isolation mechanism in honeybees. death of the participants [8]. The importance However, it might be an important factor of size and shape of the genital organs in for the positioning of a specific mating Apis shall be discussed below. period within the afternoon (see § 2.4.3).

Table III. Absolute weight of queens and drones in mg (n individuals measured).

Species Drone Queen (*virgin)

A. andreniformis 70.8 ± 3.0 (25) 112 (3) A. florea 77.6 ± 2.6 (25) 86* (5) A. cerana 83.4 ± 8.9 (38) 122 ± 13* (74) A. koschevnikovi 105.5 ± 5.6 (45) 170 (2) A. nuluensis 107.0 ± 6.7 (5) – A. nigrocincta –– A. mellifera 211.1 ± 11.8(25) 202* (5) A. dorsata 155.7 ± 8.5 (34) 290 (2) A. laboriosa – – 324 N. Koeniger, G. Koeniger

3.2. Reproductive isolation similar. The characteristic marks to distin- due to different genitalia guish them seem to be limited to differences in the hairy fields, in the form of the cor- The queen’s genital tracts within the nua and the lobe [31, 61]. During copula- genus Apis are similar and simple in princi- tion the endophallus is everted successively ple. The genital chamber opens at the base of and introduced into the queen. The differ- the sting. Its outer part is the bursa copula- ences in form and size of the everted trix, the inner part the vagina with the endophalli become striking (Fig. 4). For valvula vaginalis. Small differences con- example, the cornua bend either dorsally or cerning the valvula vaginalis have been ventrally in the everted endophallus; the reported by Camargo [7]. same is true for the bulbus [31]. These dif- ferences are expected to have major func- In comparison, differences in male gen- tional consequences in the case of copulation italia between honeybee species are very between species belonging to different tax- impressive. Generally, in Apis the copulative onomic groups (hive-dwelling, giant and organ is a membranous endophallus which dwarf honeybees). is differentiated into a broad vestibulum with cornua, a slender cervix and a thick However, within these groups there are bulb with its lobe. In situ they look quite many similarities. In all cavity-dwelling

Figure 4. Everted endophalli of nine Apis species (lateral view). Symbols for all figures: B: bulbus; Cer: cervix; P: chitinous plates of bulbus; dC: dorsal cornus; L: lobe; V: vestibulum; vC: ventral cornua. : hairy fields. Reproductive isolation among species of the genus Apis 325 species it seems to be mainly the large This probably does not hold true within endophallus which connects the drone to these groups. No difference could be found the flying queen. Filled under high muscu- between the endophalli of A. dorsata and lar pressure with mucus from the male A. laboriosa [51], thus this will not func- accessory glands and air, it seems to guar- tion as a copulatory barrier. antee a sufficiently strong connection There are some morphological differ- between the flying queen and the paralysed ences in the tip of the bulb, the fimbriate drone until sperm transfer into the oviducts lobe and the pattern of hairs between the is completed [42]. The short thick cornua dwarf species A. andreniformis and A. florea with its ‘orange coloured’ sticky and greasy (Fig. 4) [31]. Whether or not these differ- secretion may contribute to strengthen the ences are sufficient to prevent mating and attachment and later, after separation of the sperm transfer is questionable. The same pair, it sticks to the mating sign and keeps it holds true for the cavity-nesting species. in place. The giant honeybees have a more elon- 3. 3. Mating sign as a possible gated endophallus with four long curled cor- barrier nua also covered by an ‘orange coloured’ sticky secretion. The elongation of the endophallus is mainly caused by the After copulation, mating signs are left in extended cervix (Fig. 4). The drones also the sting chamber of the Apis queen. How- become paralysed during copulation. In ever, the queens of all Apis species mate these bees the broadened metatarsi seem to several times during a single mating flight. reinforce the attachment to the queen [72]. So the mating sign does not prevent further mating but it must be removed by the next The everted endophallus of the dwarf drone. In the case of interspecific mating honeybees shows many differences to the however, the endophallus of a heterospe- previous endophalli. The so-called bulbus cific drone might not be equipped to remove is a thin tube (Fig. 4), and the mucus glands the mating sign from the sting chamber. are tiny. They cannot produce enough mucus Mating signs have been described in the to strengthen the connection between the cavity-dwelling species [33, 99, 101]. They copulating pair. Instead they have a forceps- consist of mucus from the male accessory like appendix at the metatarsus of the hind glands, secretions from the bulbus gland leg, which is a bit shorter in A. andreni- and orange-coloured secretions from the formis than in A. florea. With these ‘thumbs’ cornual glands [29, 35]. In A. mellifera there the drone locks himself to the hind legs of are also chitin plates from the bulbus. In the queen [73], supported again by the sticky A. andreniformis only the secretion from cornua pressed into the sting chamber. Thus the cornual gland was found in the sting the pair stays connected until the queen turns chamber (Fig. 5) [38]. While mating signs her legs in such a way that the drone is appear similar in the Asian species A. cerana released. This different mode of attachment and A. koschevnikovi, they are clearly dif- seems to have evolved together with the ferent in A. mellifera (Fig. 5). changed form of the bulbus: it ends in a fine tip, which enables the drone to deposit the In A. mellifera it could be observed that sperm into the thin spermaduct instead of drones are able to remove the mating sign at the wide oviducts. the beginning of copulation. It is attached to the hairy field on the ventral side at the Between these three taxonomic groups − base of the endophallus. At the end of mat- dwarf, giant and cavity-dwelling honey- ing, each drone leaves its own sign [26, 96]. bees − sperm transfer very likely is pre- After the last copulation more than 70% vented by the morphology of the endophalli. mated queens return carrying the last mating 326 N. Koeniger, G. Koeniger

A. florea and A. andreniformis have tiny mucus glands. While in A. florea no secre- tion of drones was found in the sting cham- ber of queens returning from mating flights [42], all three observed queens of A. andre- niformis had a reddish yellow cornual secre- tion (Fig. 5) protruding from the tip of the abdomen [38]. The question of whether drones can remove heterospecific mating signs cannot be answered generally, but Ruttner and Maul [79] have observed that an A. cerana queen carried an A. mellifera mating sign for about 2 weeks. Dissection revealed that the chiti- nous plates had injured the membrane of the bursa copulatrix; the oviducts were filled with 1.7 µL sperm, which corresponds to one A. mellifera drone [99]. Therefore we conclude that the following A. cerana drones as well as workers (afterwards in the colony) were not able to remove the heterospecific Figure 5. Mating signs in Apis species. A.m.: (A. mellifera) mating sign. Apis mellifera ; A.c.: Apis cerana; A.k.: ; A.a.: . Mucus ; Cornual secretion ; Chitinous plates. 4. PHYSIOLOGICAL BARRIERS 4.1. Sperm transfer sign in their sting chamber. After returning The percentage of drone spermatozoa from the mating flight, this is removed stored in the spermatheca is quite different in mostly by the queen rubbing her abdomen different species, although DNA studies have on the comb [69]. As queens of A. cerana revealed that the number of effective mat- and A. koschevnikovi also return from one ings is similar among species [11, 54−58, mating flight with a mating sign and their 68]. The total number of spermatozoa in the oviducts filled with sperm of about five to 10 drones [33, 101], drones of these species, queen’s spermatheca divided by the effective too, are able to remove the mating sign of number of matings indicates the amount of their predecessors. spermatozoa contributed by each drone (Tab. IV). For example, an A. mellifera In A. dorsata, queens return from mat- drone produces about 12.7 million sperma- ing flights after sunset, during darkness! No tozoa, but only about 370 000 reach the sper- mating signs were noticed protruding from matheca. Although this is the highest num- the sting chamber, and the queen was per- ber for all species, it corresponds to only mitted to re-enter the colony [92]. On the about 3% of the spermatozoa of a single other hand, A. dorsata drones have well- drone. This number is around 10% in developed mucus glands (own observations), A. koschevnikovi, A. cerana and A. dorsata. and in a video film on mating with dum- The other extreme occurs in A. andreni- mies by Gries (unpubl. results), the depo- formis: one drone produces 0.13 million sition of a white plug on the dummy was spermatheca, of which an average of 66% demonstrated. are present in the spermatheca [30]. These Reproductive isolation among species of the genus Apis 327

Table IV. Number of spermatozoa of one drone reaching the spermatheca (million).

Species Spermatozoa in Spermatoz. Mean Effective n Spermat. Spermatoz. spermatheca in ves. sem. paternity paternity per drone in of one drone (n queens) spermatheca reaching the spermatheca (%)

A. a 0,78 (7) [38] 0.13 [30] 13.5 [57] 9.1 [57] 0.086 66 A. f 1.05 (15) [42] 0.43 [42] 8.0 [55] 5.6 [55] 0.187 44 A. c 1.35 (12) [63.101] 1.1 [63/101] 18 [58] 12 [58] 0.113 10 A. k 2.13 (4) [33 ] 1.7 [33] 16.3 [68] 10.5 [68] 0.203 12 A. m 4.73 (126) [100] 12.7 [99] 13.8 [11] 12.4 [11] 0.370 03 A. d 3.94 (8) [30, 92] 2.46 [30] 22.4 [54, 56] 22.8 [54, 56] 0.173 07

A. a: Apis andreniformis; A. f: Apis florea; A. c: Apis cerana; A. k: Apis koschevnikovi; A. m: Apis mellifera; A. d: . ( ) n queens; [ ] reference. calculations differ slightly to those of 1.9 million living spermatozoa were counted Oldroyd et al. [58] and Palmer and Oldroyd in the spermatheca of A. cerana [79]. [60], but do not alter our conclusions. In the interspecific and intraspecific insem- These findings support the idea that the inations of A. koschevnikovi and A. cerana, mode of sperm transfer (sperm injection into about the same percentage of spermatozoa − the oviducts versus sperm injection into the (8 9%) reached the spermatheca (Tab. VI), spermaduct) influences the filling process independent of hetero- or conspecific sperm. of the spermatheca. Injection into the This percentage corresponds to that after oviducts results in a loss of more than 90% natural mating (Tab. IV). In all cases, sper- spermatozoa, whereas with injection into matozoa in the spermatheca were viable the spermaduct [38] only about 50% is when queens were dissected 3 to 40 days rejected. Thus different sperm numbers after insemination. The amount of sperma- together with the different modes of sperm tozoa was below 1 million, except in A. kos- transfer may act as a partial reproductive chevnikovi queens inseminated with con- barrier. specific sperm. After insemination of A. koschevnikovi with A. dorsata sperm, the percentage of 4.2. Sperm storage spermatozoa reaching the spermatheca was quite low. But with only two experiments The technique of instrumental insemina- and no reciprocal insemination, these results tion has permitted the study of heterospecific must be considered preliminary. Woyke sperm transfer and storage. The following [102] reports that after inseminating A. flo- combinations have been made (Tab. V). rea queens instrumentally with A. mellifera sperm, some spermatozoa entered the sper- In all cases some semen reached the sper- matheca, but he did not report the percent- matheca. In the interspecific and intraspe- age. To a certain extent, a comparable cific inseminations of an A. cerana queen, migration of conspecific spermatozoa has “twice as many mellifera spermatozoa been demonstrated in A. mellifera queens reached the spermatheca when injected into [17]. No reciprocal inseminations were per- the oviducts than in the case of cerana formed. These data suggest that the physi- spermatozoa. Thus heterospecific insemi- ology of the genital duct and its fluid is sim- nation is as efficient as homospecific.” Up to ilar throughout all species. 328 N. Koeniger, G. Koeniger

Table V. Queens of 4 species instrumentally inseminated with sperm of various heterospecific species.

Sperm A. mellifera A. cerana A. koschev. A. dorsata

Queens A. mellifera + (1.2) [76, 103] A. cerana + (3) [76] + (4) [36] A. koschevnikovi + (5) [36] + (2) [36] A. florea + (7) [102]

( ) n queens; [ ] reference.

Table VI. Number of spermatozoa (million) in queens after instrumental insemination.

cer × cer cer × kosch kosch × kosch kosch × cer kosch × dors (15 queens) (10 queens) (4 queens) (4 queens) (2 queens) (%) (%) (%) (%) (%) n Drones 8 ± 1.8 7 ± 1.9 9 ± 17 ± 1 5 n Sperm in oviducts 7.3 ± 1.4 8.8 ± 2.4 11.7 ± 1.8 6.4 ± 1.9 9.0 n Sperm in spermatheca 0.7 ± 0.3 0.7 ± 0.3 1.1 ± 0.3 0.6 ± 0.2 0.2 / 0.5 % Sperm in spermatheca 9.2 ± 3.2 8.1 ± 4.4 9.3 ± 1.7 9.1 ± 2.2 2.2 / 5.6

5. POSTZYGOTIC BARRIER matozoa, however, showed patterns which at least were similar to the arrangement of an 5.1. Fertilisation and hybrids embryo [79]. With these results, complete reproductive isolation was proven between Ruttner and Maul [79] collected eggs of these two species. cross-inseminated queens of both A. cerana In 1996, five A. cerana queens insemi- and A. mellifera 1 hour after deposition. In nated with A. koschevnikovi sperm produced 92% of the eggs “motile spermatozoa with progeny. Brood counts revealed altogether rapidly undulation movements” could be 338 drone and 454 worker cappings detected close to the anterior pole. But no (Tab. VII). These queens had 0.6 ± 0.3 mil- larvae developed. Squash preparations from lion spermatozoa in the spermatheca. In the late cleavage stages, fixed and stained, brood of two queens, we found both drone showed normal diploid chromosome sets at heads and worker heads. Because of the same rates. Serial sections revealed that unfavourable circumstances, further analy- hybrid eggs of both types could form a ‘pre- sis was not possible. In 1997, sealed brood blastoderm’, but during the blastula stage combs of two A. cerana queens inseminated the initial cell walls disintegrated again and with A. koschevnikovi sperm were kept in development ended with a complete break- the incubator and some hybrids with gynan- down. Apparently there is some variation dromorph characters were reared (Tab. VIII). in the degree of embryonic disintegration. From one queen we collected 10 drones and Reciprocal hybrids with A. cerana from Pak- 14 hybrids. In 13 worker-like hybrids, the istan were all highly disintegrated by the distance between the complex eyes was third day of development. Eggs of A. mel- more than 1.5 mm; in one bee, this measured lifera queens fertilised by A. cerana sper- only 0.15 mm. All had a long proboscis, Reproductive isolation among species of the genus Apis 329

11 hybrids had no stings, and except for three hybrids) between A. cerana and A. mellifera bees which had 1 or 2 drone hind legs, all [79]. The question of the taxonomic status of the others had worker legs. From the other A. cerana as a valid species was settled queen we collected 24 offspring: 10 drones (Tab. IX). and 14 that appeared to be drone-like The next step in research on reproduc- hybrids, which were recognised because of tive isolation was the discovery by Koeniger the dorsal distance between the complex and Wijayagunesekera [39] that different eyes. The distances varied between 0.20 and daily mating periods served as complete 1.55 mm; in 11 hybrids, the distance was mating barriers among sympatric A. florea, below 1.0 mm. The ocelli were situated A. cerana and A. dorsata in Sri Lanka. These either frontally or between the complex eyes. results laid the basis for further develop- Six hybrids had short drone and eight had ments. In 1988, Ruttner [73] discussed his long worker proboscises. All but one had research on A. cerana and A. mellifera and drone legs. In further experiments we will argued “that no premating barrier exists try to produce hybrid queens and test their between these two species as is the case ability to reproduce. No data on hybrids between the other species. Because of a not between A. koschevnikovi queens insemi- completely finished speciation, it appears nated with A. cerana sperm are yet avail- totally unjustified to classify this taxon able. (A. cerana) as a subgenus, as was proposed It seems unlikely that the above hybrid by Skorikov [89] and Maa [49]. On the con- bees could form a viable colony. Though trary, they (A. mellifera and A. cerana) have we have not yet tried to breed hybrid queens, to be regarded as being in a late, but not yet the results suggest complete reproductive finished stage of speciation” ([73] p. 148). isolation. According to the DNA and mor- So the ‘missing premating barrier’ between phological results [2, 12], A. cerana is more species was used as a basis for determination related to A. nuluensis than to A. koschev- of a taxonomic position. We will postpone nikovi. So cross-insemination between these commenting on that argument, and first pur- species may result in ‘better’ hybrid bees. sue the role of ‘reproductive isolation research’ in the recognition of further ‘new’ Apis species. 6. CONCLUSION The rediscovery of A. koschevnikovi in Sabah and its recognition as a valid species Evidence of reproductive isolation among was based on a combination of different distinct honeybee populations has been used results. The first argument was sympatric for confirmation and recognition of new distribution with A. cerana, then morpho- species and for estimation of taxonomic rela- metric differences between both species and tions. We now discuss the value and the lim- distinct differences in the morphology of the itations of this approach. endophallus were demonstrated [94]. After this, evidence for complete reproductive iso- In the 1960s there were generally only lation by a separated daily mating period led four recognised species in the genus. Even to the ‘final’ acceptance of A. koschevnikovi the discussion about the taxonomic status as a ‘good species [41]. Since then, deter- of the two cavity-dwelling bees, A. mellifera mination of mating periods has become and A. cerana, was not yet settled. In this indispensable for the recognition of ‘new’ context, Ruttner and co-workers started their honeybee species. first research project on reproductive isola- tion. As a result, a full array of post-mating Research on post-mating barriers (see § 5) barriers was presented, culminating in evi- between A. cerana and A. koschevnikovi dence of genetic incompatibility (lethal revealed similarities. In all combinations 330 N. Koeniger, G. Koeniger

Table VII. 1996: Progeny of A. cerana queen inseminated with A. koschevnikovi sperm from one brood comb and numbers of spermatozoa in the spermatheca.

Insem. Eggs Larvae Dronecap Workercap n spermat. (million)

1. A. c. × 8 A. k. 38 22 55 16 0.3 2 A. c. × 5 A. k. 86 43 33 64 1.2 3. A. c. × 6 A. k. 68 60 94 6 0.3 4. A. c. × 6 A. k. ? ? 39 121 0.3 5 A. c. × 8 A. k. 132 263 117 247 0.7

Queens 4 and 5 already had pupae; drone heads and worker heads could be distinguished. Insem.: insemination; spermat.: spermatozoa.

Table VIII. Offspring of two A. cerana queens inseminated with A. koschevnikovi spermatozoa.

Character Workerlike Mix Drone like

Qu 1 Qu 2 Qu 1 Qu 2 Qu 1 Qu 2

Eye distance 11 3 0 9 1 12 Proboscis 12 7 0 1 0 16 Tomenta 1 6 6 0 0 6 18 Leg 19 0 1 1 2 13 Sternite 11 0 3 0 8 14

Qu: queen.

Table IX. Taxonomic distance and mode of reproductive isolation between cavity dwelling honey- bee species.

Species to species Taxonomic Natural Premating Postmating distance distribution barrier barrier A. cerana – A. nuluensis Nearest Sympatric, Mating period Not known different habitat A. cerana – A. koschevnikovi Second Sympatric, Mating periodEndophallus same habitat viable hybrids A. cerana – A. mellifera Third Allopatric None Endophallus lethal hybrids

among the three cavity-dwelling species, nikovi were viable and were reared to adult heterospecific spermatozoa reached and workers (with some gynandromorph pat- stayed viable in the spermatheca. There terns). were, however, some remarkable differ- The next ‘case’ to be discussed was the ences: while hybrids between A. cerana and discovery of A. nuluensis. As already men- A. mellifera died during blastogenesis, tioned, the habitat of A. nuluensis is hybrids between A. cerana and A. koschev- restricted to the mountains of Borneo where Reproductive isolation among species of the genus Apis 331 it is the only honeybee species. The main fect behavioural mating barriers. argument for the recognition of the new A. mellifera is the sister taxon to the species was again evidence that the drone Asian cavity-dwelling species, and its sep- flight period of A. nuluensis was different aration must have occurred significantly ear- from that of the other cavity-nesting species lier than the above branchings [87]. There- of Borneo (A. cerana and A. koschevnikovi). fore, the ‘missing’ premating barrier does No data on post-mating barriers are yet not correlate to a recent speciation. It is more available. Last, the recognition of Apis likely that because of the (still existing) nigrocincta from as a separate allopatric situation of A. mellifera, there was species was mainly based on differences no adaptive pressure for the development between mating periods to those of the sym- of any type of premating barrier. In other patric A. cerana [22]. words, the evolution of a behavioural mat- Collectively, the group of cavity-dwelling ing barrier in honeybees seems to be con- honeybees today comprises five species. nected to a sympatric co-existence and does This offers a new and promising base to not depend on the taxonomic position. In explore the value of data on reproductive contrast, non behavioural mating barriers isolation for determination of the taxonomic seem to be a result of the general genetic relations among Apis species. For ease of divergence, which starts with the reduction discussion and because several comparable and eventual elimination of gene flow taxonomic papers do not include it, we have between the two populations. Therefore the omitted the very recently recognised degree of differences in endophalli, in via- A. nigrocincta from the following consid- bility of hybrids etc. (see B and C) corre- erations. lates well with the phylogenetic distance. As a basis for the taxonomic positions of the species, we refer to the consistent results of several morphometric studies [12, 66, 82] ACKNOWLEDGMENTS which can be summarised in terms of a hierarchy of taxonomic distances: A. cer- The enjoyable and productive long-term coop- eration with Salim Tingek and his team at A.R.S. ana (Borneo) / A. nuluensis < A. cerana / Tenom was substantial for many of the above- A. koschevnikovi < A. cerana / A. mellifera. reported results. Further, we wish to thank Supporting genetic evidence for the above Mr. Matthew Tulin Tulas, the Principal Research sequence comes from Arias et al. [2], and Officer and the Officer in Charge of the Agri- parts of it from Garnery et al. [14]. cultural Research Station Tenom for invaluable assistance and support. Dr. Stefan Fuchs has read The nearest taxonomic distance within and discussed earlier versions of the manuscript, the above group and probably the most and we would like to thank him for several con- recent speciation separated A. cerana (Bor- structive suggestions. Last but not least, we owe neo) and A. nuluensis. The species are still special thanks to Steve Sheppard who spent many (?) distributed in ecologically different habi- hours improving the English linguistics and style. tats. The behavioural barrier might have an adaptive value in a small zone of overlap (1 500 to 1 700 m above sea level) with the Résumé – L’isolement reproductif parmi other cavity-nesting species. However, les espèces du genre Apis. Les études sur between A. cerana and A. koschevnikovi the l’isolement reproductif des abeilles du genre separated daily mating periods are of con- Apis ont commencé relativement tard par siderable adaptive significance because both rapport aux autres animaux ou aux plantes. species share the same habitat. So the three Cela est dû principalement au fait que les cavity-dwelling Apis species of Borneo (tax- colonies d’abeilles qui étaient importées onomically close to each other) have per- d’Afrique ou du Bassin méditerranéen vers 332 N. Koeniger, G. Koeniger l’Europe appartenaient à la même espèce barrière reproductive a été étudiée à l’aide de (Apis mellifera L.), et se croisaient donc leurres (Fig. 2). On a pu montrer que les avec les abeilles déjà présentes. Ce n’est mâles pouvaient être attirés sur un LRM qu’en 1965 que F. Ruttner, après avoir d’une espèce étrangère et y copuler avec importé de Chine des colonies d’A. cerana des leurres spéciaux (Fig. 3). Les différents Fabricius, put ouvrir un nouveau chapitre LRM ne peuvent donc pas jouer pleinement de la recherche apidologique par des tra- le rôle de barrière reproductive entre espèces vaux expérimentaux sur les barrières repro- du genre Apis. Les mâles réagissent aux ductives entre races d’abeilles. extraits de reines d’espèces étrangères, et Traditionnellement les barrières reproduc- les mâles de différentes espèces copulent tives ont été classées selon un ordre chro- avec un même leurre, imprégné que d’une nologique par rapport à la fécondation en phéromone (9-ODA) (Fig. 3). On ne peut obstacles prézygotiques et postzygotiques. donc attribuer qu’une importance nulle ou Afin de mieux prendre en compte les faible au bouquet phéromonal des reines en particularités biologiques du processus com- tant que barrière reproductive. pliqué de l’accouplement multiple chez le Des périodes d’accouplement différentes au genre Apis, nous avons divisé la phase pré- cours de la journée viennent en première zygotique en trois sous-groupes (Tab. I) et position parmi les barrières reproductives utilisé les définitions suivantes : liées au comportement. Le tableau II résume Barrières comportementales. – Les barrières les résultats des recherches sur ce point et comportementales empêchent le contact montre des différences dans les périodes physique entre la reine et les mâles. Les bar- d’accouplement au sein d’une même espèce. rières à la copulation interviennent pendant Néanmoins, la succession chronologique la copulation et empêchent le transfert du des périodes d’accouplement spécifiques à sperme. Les barrières physiologiques blo- chaque espèce est dans l’ensemble sem- quent la voie au sperme entre les oviductes blable au Sri Lanka, en Thaïlande et à Bor- de la reine et la spermathèque, et plus tard la néo. Les abeilles naines (A. andreniformis pénétration dans l’œuf. Les barrières post- et/ou A. florea) arrivent en première posi- zygotiques perturbent le développement nor- tion, au moment où le soleil est au zénith, ou mal et conduisent à la mort ou à l’infertilité juste après. La période suivante est occu- des hybrides. pée par les abeilles qui nidifient dans des cavités. Et peu avant le coucher du soleil, Barrières reproductives liées au comporte- les mâles d’A. dorsata sortent pour effec- ment. – La formation des essaims de multi- tuer leur bref vol nuptial. Il se dégage de plication et les accouplements n’ont lieu en l’ensemble un schéma uniforme, et la posi- règle générale qu’à la belle saison. En consé- tion de la période d’accouplement propre à quence, les espèces sympatriques produi- une espèce semble n’être corrélée qu’avec la sent des mâles et des reines toujours en taille de l’insecte, que ce soit entre groupes même temps ; il n’y a donc pas de diver- taxonomiques ou au sein d’un même groupe. gence dans la saison d’accouplement qui Il est encore prématuré de spéculer sur les puisse agir comme barrière reproductive mécanismes à l’origine de cette succession entre espèce d’Apis. chronologique, mais une particularité du Un schéma général des lieux de rassemble- comportement d’accouplement des mâles ments de mâles (LRM) de trois espèces pourrait jouer un rôle. La réaction des mâles (Fig. 1) a été déduit des observations faites est dirigée vers les reines qui, chez toutes en trois lieux différents à Bornéo. La ques- les espèces du genre Apis, sont plus grosses tion de savoir dans quelle mesure la sépa- que les mâles. Aussi les mâles des plus petits ration spatiale trouvée entre les LRM spé- espèces pourraient essayer de copuler avec cifiques à chaque espèce est valable comme les mâles des plus grosses en les prenant Reproductive isolation among species of the genus Apis 333 pour des reines et les empêcher ainsi d’avoir était resté enfoncé dans l’oviducte. Visible- accès aux reines. Mais de telles considéra- ment le signe de fécondation de l’espèce tions ne reposent pas encore sur des bases étrangère n’avait pu être ôté. Cela pourrait expérimentales, et il est tout autant possible être valable pour des accouplements entre que ce ne soit pas la taille corporelle elle- autres espèces car les signes de fécondation même qui soit décisive, mais des facteurs sont construits différemment (Fig. 5). physiologiques liés à la taille, pour que les Barrières reproductives liées à la physiolo- espèces les plus petites volent en premier. gie. – La variation selon l’espèce du nombre Notre hypothèse en est arrivée au test à de spermatozoïdes des mâles semble empê- grande échelle, puisque A. florea a été récem- cher une insémination suffisante. Ainsi une ment introduite accidentellement en Afrique, reine d’A. mellifera devrait s’accoupler avec où ses populations s’étendent. Elles vont 120 mâles d’A. cerana pour avoir une quan- peut-être atteindre l’aire de répartition tité de sperme équivalent à celui transféré naturelle d’A. mellifera scutellata et parve- par 12 mâles d’A. mellifera. Le tableau IV nir à une coexistence sympatrique. Dans ce donne le nombre de spermatozoïdes trans- cas, nous prédisons que les mâles d’A. florea féré par un mâle de chaque espèce. Le pour- accompliront leurs vols de fécondation avant centage de spermatozoïdes qui atteint la ceux d’A. mellifera. spermathèque varie également d’une espèce à l’autre (Tab. IV). Barrières reproductives liées à la copula- La physiologie du tractus génital et de la tion. – Des différences de taille aussi bien spermathèque semble par contre semblable. que des différences de structure des organes Après insémination artificielle, les sperma- copulateurs peuvent empêcher la copula- tozoïdes sont parvenus jusqu’à présent dans tion. Les deux situations se rencontrent chez la spermathèque dans tous les cas et sont les abeilles naines et les abeilles géantes. restés viables durant la période testée (de Les abeilles qui nidifient dans les cavités se trois jours à cinq semaines) (Tabs. VI et distinguent elles aussi nettement des espèces VII). sus-mentionnées. Mais au sein des groupes ce type de barrières n’est pas total. Les Barrières postzygotiques. – Des essais de organes de copulation des abeilles géantes croisement entre espèces n’ont été menés à A. laboriosa et A. dorsata ont la même struc- ce jour que sur une échelle réduite. Lors de ture et les rapports de taille sont semblables. croisements dans les deux sens entre Chez les deux espèces d’abeilles naines A. cerana et A. mellifera, des œufs ont bien (A. florea, A. andreniformis) les organes de été fécondés, mais l’embryon est mort 48 h copulation, bien qu’ayant des différences plus tard pendant la formation de la blas- nettes dans la pilosité et la forme, sont bâtis tula. Lors de croisements entre deux reines sur le même principe, si bien qu’une copu- d’A. cerana avec des mâles d’A. koschev- lation semblerait possible. Mais puisque les nikovi, quelques hybrides se sont développés sexués volent à des heures différentes, il n’y et ont donné des ouvrières. Une reine a a aucune possibilité de le tester. Cela vaut donné naissance à des gynandromophes également pour toutes les espèces sympa- ayant un aspect de mâles, que l’on pouvait triques. L’importation d’A. cerana en Europe reconnaître principalement au grand écar- a permis d’étudier les interactions interspé- tement de leurs yeux (Tab. VIII). Les autres cifiques lors de l’accouplement. Au début hybrides possédaient de nombreuses carac- de nombreuses reines d’A. cerana ont été téristiques des ouvrières. perdues. Une reine d’A. cerana, qui était En conclusion, nous discutons le rôle qu’a rentrée du vol de fécondation mais n’avait joué la recherche sur les barrières repro- pas pondu d’œufs, a été mutilée par le signe ductives dans la reconnaissance de nouvelles de fécondation d’un mâle d’A. mellifera qui espèces d’abeilles et dans la détermination 334 N. Koeniger, G. Koeniger des distances taxonomiques. La démons- die Spermatheka und weiter bis zum Ein- tration des différentes périodes de vol des dringen ins Ei. Postzygotische Barrieren mâles a été d’une importance décisive dans stören die normale Entwicklung und führen la reconnaissance d’A. koschevnikovi, zum Tod bzw. zur Unfruchtbarkeit der d’A. nuluensis et d’A. nigrocincta. Par contre, Hybriden. les « barrières post-accouplement », tels que Verhaltensbedingte Kreuzungsbarrieren. – la forme de l’endophallus, le transfert de Die Bildung von Vermehrungsschwärmen sperme et la viabilité des hybrides, convien- und auch die Paarungen finden in der Regel nent mieux pour estimer la distance taxo- nur in der günstigsten Jahreszeit statt. Ent- nomique. sprechend erzeugen sympatrische Arten stets zur gleichen Jahreszeit Drohnen und Köni- Apis / isolement reproducteur / compor- ginnen und eine unterschiedliche Paa- tement d’accouplement / genitalia / rungssaison als Kreuzungsbarriere zwischen hybride Apisarten scheidet weitgehend aus. Ein allgemeines Schema der Drohnensam- melplätze von drei Arten (Abb. 1) wurde Zusammenfassung – Kreuzungsbarrie- von Ergebnissen an drei unterschiedlichen ren zwischen Arten der Gattung Apis. Plätzen in Borneo abgeleitet. Die Frage Forschungen über reproduktive Isolation inwieweit die gefundene räumliche Tren- von Honigbienen begannen im Vergleich nung zwischen den artspezifischen Droh- zu anderen Tieren oder Pflanzen relativ spät. nensammelplätzen als Paarungsschranke Das lag vor allem daran, daβ alle Bienen- gelten kann, wurde mit Attrappen (Abb. 2) völker, die aus Afrika oder dem Mittel- untersucht. Es konnte gezeigt werden, daβ meerraum nach Europa importiert wurden, Drohnen auf artfremden Drohnensammel- zur gleichen Art (Apis mellifera L.) gehör- plätzen angelockt werden konnten und dort ten und daher mit den bereits vorhandenen mit speziellen Attrappen (Abb. 3) kopulier- Bienen bastardierten. Erst 1965 konnte ten. Demnach können die verschiedenen F. Ruttner nach der Einfuhr von A. cerana Paarungsplätze nicht als vollständige Kreu- Bienenvölkern aus China mit experimen- zungsbarriere zwischen den Honigbienen- tellen Arbeiten über Kreuzungsbarrieren arten gelten. Weiter wird diskutiert, inwie- zwischen Honigbienenarten dieses neue weit unterschiedliche Königinnenpheromone Kapitel der Bienenforschung aufschlagen. als Kreuzungsbarriere dienen können. Droh- Traditionell werden Kreuzungsbarrieren nen reagieren auf Extrakte von artfremden nach ihrer zeitlichen Zuordnung zur Königinnen und – soweit bisher getestet – Befruchtung in präzygotische oder postzy- kopulieren die Drohnen verschiedener Arten gotische Barrieren eingeteilt. Um die Beson- mit einer identischen Attrappe, die nur mit derheiten der komplizierten Paarungsbiolo- einem Pheromon (9-ODA) kontaminiert war gie der Apisarten besser zu berücksichtigen, (Abb. 3). Demnach scheint den unter- haben wir die präzygotische Phase noch ein- schiedlichen Pheromonmustern der Köni- mal in drei Untergruppen geteilt (Tab. I). ginnen keine oder eine nur geringe Bedeu- Für diese Arbeit gelten folgende Definitio- tung als Kreuzungsbarriere zu zukommen. nen: Als verhaltensbedingte Kreuzungsbarriere Verhaltensbedingte Kreuzungsbarrieren ver- kommen in erster Linie tageszeitlich unter- hindern den Kontakt zwischen Königin und schiedliche Paarungsperioden in Frage. In Drohnen. Kopulationsschranken greifen Tabelle II sind Ergebnisse von Untersu- während der Kopulation ein und verhindern chungen zusammengefaβt. Dabei ergaben die Übertragung von Spermien. Physiolo- sich Unterschiede zwischen den Paarungs- gische Barrieren blockieren den Weg der zeiten innerhalb einer Art. Insgesamt war Spermien von den Ovidukten der Königin in jedoch die zeitliche Folge der artspezifischen Reproductive isolation among species of the genus Apis 335

Paarungsperioden in Sri Lanka, Thailand innerhalb der Gruppen sind derartige Bar- und Borneo gleich. Die Zwerghonigbienen rieren sicher nicht vollständig. Die Begat- (A. andreniformis und/oder A. florea) halten tungsorgane von den Riesenhonigbienen die erste Position nahe bzw. kurz nach dem A. laboriosa und A. dorsata sind gleich Zenit der Sonne. Die nächste Paarungsperi- gebaut und auch die Gröβenverhältnisse ode wird von höhlenbrütenden Arten besetzt. sind ähnlich. Bei den beiden Arten der Kurz vor Sonnenuntergang starten die Droh- Zwerghonigbiene sind die Begattungsor- nen von A. dorsata zu ihrem kurzen Paa- gane trotz deutlicher Unterschiede in der rungsflug. Insgesamt entsteht so ein ein- Behaarung und Form nach dem gleichen heitliches Muster und die Position der Prinzip geformt, so daβ eine Paarung mög- artlichen Paarungsperiode scheint zwischen lich erschiene. Da die Geschlechtstiere zu und auch innerhalb der taxonomischen unterschiedlichen Zeiten fliegen, gibt es Gruppe allein mit der Gröβe zu korrelieren. keine Möglichkeiten, dies zu testen. Glei- Es ist noch voreilig zu spekulieren, auf ches gilt für alle sympatrischen Arten. Der Grund welcher Mechanismen diese zeitli- Import von A. cerana nach Europa ermög- che Reihenfolge entstanden ist. Allerdings lichte Untersuchungen über zwischenartliche könnte eine Besonderheit des Paarungsver- Interaktionen bei der Paarung. Zunächst haltens der Drohnen dabei von Bedeutung waren die Verluste von A. cerana Königin- sein. Die Reaktion der Drohnen ist auf Köni- nen hoch. Eine A. cerana Königin, die zwar ginnen gerichtet, die bei allen Honigbie- zurückkehrte aber keine Eier legte, war nenarten grundsätzlich gröβer sind als die durch das im Ovidukt steckengebliebene Drohnen. So könnten die Drohnen der grö- Begattungszeichen eines A. mellifera Drohns βeren Art “für Königinnen gehalten wer- verletzt. Offensichtlich konnte das Begat- den” und gezielt von Drohnen der kleine- tungszeichen der anderen Art nicht entfernt ren Art angeflogen und auf diese Weise werden. Das könnte auch für Paarungen zwi- verdrängt werden. Aber solche Überlegun- schen anderen Arten zutreffen, denn die gen sind noch nicht experimentell gesichert Begattungszeichen sind unterschiedlich und es ist gleichfalls möglich, dass nicht die gebaut (Abb. 5). Köpergröβe direkt sondern physiologische Faktoren, die mit der Gröβe korrelieren. Physiologische Barrieren. – Die unter- Entscheidend dafür sind, dass die kleineren schiedlichen Spermazahlen der Drohnen Arten früher fliegen. Insgesamt steht der sprechen für eine Behinderung einer aus- Hypothese, dass jeweils die kleineren Droh- reichenden Besamung bei verschiedenen nen früher fliegen, ein groβer Freilandtest Arten. So müβte eine A. mellifera Königin bevor. Es ist zu erwarten, dass sich A. florea von 120 A. cerana Drohnen gepaart wer- weiter in Afrika ausbreitet und im natürli- den, um die gleiche Menge Spermien zu chen Verbreitungsgebiet von A. m. scutellata erhalten die 12 A. mellifera Drohnen über- kann es zu einer sympatrischen Koexistenz tragen. Die Unterschiede in den Sperma- kommen. Wir sagen voraus, dass dann die zahlen sind in Tabelle IV angegeben. Unter- A. florea vor den Drohnen von A. mellifera schiedlich sind auch die Prozentsätze von zur Paarung ausfliegen werden. Spermien, die bei den Arten in die Sperma- theka gelangten (Tab. IV). Kopulationsbarrieren. – Sowohl unter- schiedliche Gröβe als auch unterschiedlich Die Physiologie des Genitaltraktes und der gebaute Begattungsorgane können Kopula- Spermatheka dagegen scheint ähnlich zu tionen verhindern. Bei den Bienen trifft bei- sein. Nach künstlicher Besamung gelangten des für die Zwerg- und Riesenhonigbiene in allen bis jetzt getesteten Fällen (Tab. VI zu. Auch die höhlenbrütenden Arten unter- und VII) Spermien in die Spermatheka und scheiden sich deutlich von den vorgenann- waren für die getestete Zeit (3 Tage bis ten Zwerg – bzw. Riesenhonigbienen. Aber 5 Wochen) lebensfähig. 336 N. Koeniger, G. Koeniger

Postzygotische Barrieren. – Versuche zu [4] Butler C.G., Fairey E.M., Pheromones of the Hybridisierungen sind bisher nur in gerin- honeybee: biological studies of the mandibular gland secretion of the queen, J. Apic. Res. 3 gem Umfang durchgeführt worden. Bei (1964) 65–76. Kreuzungen von A. cerana mal A. mellifera [5] Butler C.G., Calam D.H., Callow R.K., Attrac- in beiden Richtungen wurden Eier zwar tion of Apis mellifera drones by the odours of befruchtet, aber der Embryo starb nach 48 the queens of two species of honeybees, Nature 213 (1967) 423–424. Stunden während der Bildung der Blastula. [6] Callow R.K., Johnston N.C., The chemical con- Bei der Kreuzung von zwei A. cerana Köni- stitution and synthesis of queen substances of ginnen mit A. koschevnikovi Drohnen ent- honeybees (Apis mellifera L.), Bee World 41 wickelten sich einige Hybriden bis zu ferti- (1960) 152–153. gen Arbeiterinnen. Bei einer Königin [7] Camargo J.M.F., Notas prévias sobre o trato gen- tal de rainhas de Apis dorsata e Apis florea entstanden drohnenähnliche Gynandromor- (Hym.: ) Homenagem à Warwick E. Kerr, phe, die vor allem an ihrem groβen Augen- Rio Claro (1972) 47–55. abstand zu erkennen waren (Tab. VIII). Die [8] Dobzhansky T., Ayala F.J., Stebbins G.L., anderen Hybriden hatten viele Eigenschaf- Valentine J.W., Evolution, W.H. Freeman and ten von Arbeiterinnen. Co., San Francisco, CA, 1977. β [9 ] Drescher W., Die Flugaktivität von Drohnen der Abschlie end wird diskutiert, welche Rolle Rasse Apis mellifera carnica und Apis mellifera die Forschung über Kreuzungsbarrieren bei ligustica in Abhängigkeit von Lebensalter und der Anerkennung neuer Honigbienenarten Witterung, Z. Bienenforsch. 9 (1969) 390–409. bzw. bei der Bestimmung von taxonomi- [10] Dzierzon J., Theorie und Praxis des neuen schen Distanzen gespielt hat. Dabei war vor Bienenfreundes, (publ. by the author), 1848. [11] Estoup A., Solignac M., Cornuet J.-M., Precise allem der Nachweis von unterschiedlichen assessment of the number of patrilines and of Drohnenflugzeiten für die Anerkennung genetic relatedness in colonies, Proc. von A. koschevnikovi, A. nuluensis und R. Soc. Lond. Ser. B Biol. Sci. 258 (1994) 1–7. A. nigrocincta von entscheidender Bedeu- [12] Fuchs S., Koeniger N., Tingek S., The morpho- metric position of Apis nuluensis (Tingek, tung. Für die Abschätzung der systemati- Koeniger and Koeniger, 1996) within cavity- schen Distanz dagegen sind offensichtlich nesting honey bees, Apidologie 27 (1996) “post mating barriers”, w.z.B. Gestalt des 397–406. Endophallus, Spermaübertragung und [13] Fujiwara S., Miura H., Kumagai T., Sawaguchi T., Lebensfähigkeit von Hybriden aussage- Naya S., Goto K.T., Suzuki K., Drone congre- gation of in an open area kräftiger. over larger trees (Zelkovia serrata), Apidologie 25 (1994) 331–337. Apis / Kreuzungsbarriere / Paarungsver- [14] Garnery L., Vautrin D., Solignac M., Phyloge- halten / Paarungsorgan / Hybrid netic relationships in the genus Apis inferred from mitochondrial DNA sequence data, Api- dologie 22 (1991) 87–92. [15] Gary N.E., Chemical mating attractants in the REFERENCES queen honeybee, Science 136 (1962) 773–774. [16] Gary N.E., Marston J., Mating behaviour of [1] Alcock J., Body size and its effect on male–male drone honey bees with queen models (Apis mel- competition in Hylaeus alcyoneus (Hyme- lifera L.), Anim. Behav. 19 (1971) 299–304. nopterea: Collitidae), J. Insect Behav. 4 (1995) [17] Gessner B., Ruttner F., Transfer der Spermato- 149–159. zoen in die Spermatheka der Bienenkönigin, [2] Arias M.C., Tingek S., Kelitu A., Sheppard W.S., Apidologie 8 (1977) 1–18. Apis nuluensis Tingek, Koeniger and Koeniger, [18] Gries M., Vergleichende Untersuchungen zum 1996 and its genetic relationship with sympatric Flugverhalten von Drohnen der Gattung Apis species inferred from DNA sequences, Api- bei der Königinnenverfolgung, Dissertation am dologie 27 (1996) 415–422. Fachbereich Biologie der Universität Frankfurt [3] Baudry E., Solignac M., Garnery L., Gries M., am Main, 1997. Cornuet J.M., Koeniger N., Relatedness among [19] Gries M., Koeniger N., Straight forward to the honeybees (Apis mellifera ) of a drone congre- queen: pursuing honeybee drones (Apis mellifera gation, Proc. R. Soc. Lond. B 265 (1998) L.) adjust their body axis to the direction of the 2009–2014. queen, J. Comp. Physiol. A 179 (1996) 539–544. Reproductive isolation among species of the genus Apis 337

[20] Haas A., Arttypische Flugbahnen von Hum- [35] Koeniger G., Koeniger N., Tingek S., Interspe- melmännchen, Z. Vgl. Physiol. 31 (1949) cific instrumental insemination of Apis cerana 281–307. Fabricius, 1793, Apis koschevnikovi v. Buttel- [21] Hadisoesilo S., A comparative study of two Reepen, 1906 and Apis dorsata Fabricius, 1793, species of cavity nesting honey bees of Sulawesi, Apidologie 27 (1996) 303–304. , Ph.D. thesis, Univ. Guelph, Ontario, [36] Koeniger G., Hänel H., Wissel M., Herth W., Canada, 1997. Cornual gland of the honey beedrone (Apis mel- [22] Hadisoesilo S., Otis G.W., Drone flight times lifera L.): structure and secretion, Apidologie confirm the species status of 27 (1996) 145–156. Smith, 1961 to be a species distinct from Apis [37] Koeniger G., Koeniger N., Tingek S., Kelitu A., cerana F., 1793 in Sulawesi, Indonesia, Api- Preliminary report on hybrids between Apis cer- dologie 27 (1996) 361–369. ana Fabricius, 1793 and Apis koschevnikovi v. [23] Hennig W., Die Stammesgeschichte der Insekten, Buttel-Reepen, 1906 produced by instrumental Verlag W. Kramer, Frankfurt am Main, 1969. insemination, Apidologie 29 (1998) 462–463. [24] Jean-Prost P., Observation sur le vol nuptial des [38] Koeniger G., Koeniger N., Tingek S., Kelitu A., reines d’abeilles, C.R. Acad. Sci. Paris 245 Mating flights and sperm transfer in the dwarf (1957) 2107–2110. honey bee Apis andreniformis (Smith, 1858), [25] Kerr W.E., Bueno D., Natural crossing between Apidologie 31 (2000) 301–311. Apis mellifera adansonii and Apis mellifera ligus- tica, Evolution 24 (1970) 145–148. [39] Koeniger N., Wijayagunesekera H.N.P., Time of drone flight in the three Asian honeybee [26] Koeniger G., Mating sign and multiple mating in species (Apis cerana, Apis florea, Apis dorsata), the honeybee, Bee World 67 (1986) 141–150. J. Apic. Res. 15 (1976) 67–71. [27] Koeniger G., The role of the mating sign in [40] Koeniger N., Koeniger G., An evolutionary honey bees Apis mellifera L.: does it hinder or approach to mating behavior and drone copula- promote multiple mating? Anim. Behav. 39 tory organs in Apis, Apidologie 22 (1991) (1990) 444–449. 581–590. [28] Koeniger G., Hänel H., The bulbus gland of drones (Apis mellifera L.), Pszcz. Zesz. Nauk. [41] Koeniger N., Koeniger G., Tingek S., Mardan XL (1996) 45–54. M., Rinderer T.E., Reproductive isolation by different time of drone flight between Apis cer- [29] Koeniger G., Koeniger N., Pechhacker H., ana Fabricius, 1793 and Apis vechti Maa, 1953, Ruttner F., Berg S., Assortative mating in a Apidologie 19 (1988) 103–106. mixed population of European honeybees, Apis mellifera ligustica and Apis mellifera carnica, [42] Koeniger N., Koeniger G., Wongsiri S., Mating Insectes Soc. 36 (1989) 129–138. and sperm transfer in Apis florea, Apidologie [30] Koeniger G., Koeniger N., Mardan M., Otis 20 (1989) 413–418. G.W., Punchihewa R.W.K., Otis G.W., Num- [43] Koeniger N., Koeniger G., Tingek S., Kelitu A., bers of spermatozoa in queens and drones indi- Mardan M., Drones of Apis dorsata Fabricius, cate multiple mating in Apis andreniformis and 1793 congregate under the canopy of tall emer- Apis dorsata, Apidologie 21 (1990) 281–186. gent trees in Borneo, Apidologie 25 (1994) [31] Koeniger G., Koeniger N., Mardan M., Otis 249–264. G.W., Wongsiri S., Comparative anatomy of [44] Koeniger N., Koeniger G., Tingek S., Kelitu A., male genital organs in the genus Apis, Apidolo- Interspecific rearing and acceptance of queens gie 22 (1991) 539–552. between Apis cerana Fabr. 1793 and Apis [32] Koeniger G., Koeniger N., Mardan M., Wongsiri koschevnikovi Buttel-Reepen 1906, Apidologie S., Variance in weight of sexuals and workers 27 (1996) 371–380. within and between 4 Apis species (Apis florea, [45] Koeniger N., Koeniger G., Gries M., Tingek S., Apis dorsata, Apis cerana and Apis mellifera), in: Kelitu A., Reproductive isolation of Apis nulu- Asian Apic; Proc. 1st Int. Conf. Asian Honey ensis Tingek, Koeniger and Koeniger, 1996 by Bees and Bee , Bangkok, Thailand. Wicwas Press, Cheshire, CT, USA, 1993, pp. 104–109. species-specific mating time, Apidologie 27 (1996) 353–359. [33] Koeniger G., Koeniger N., Tingek S., Mating flights, number of spermatozoa, sperm transfer [46] Koeniger N., Tingek S., Koeniger G., Gries M., and degree of in Apis koschevnikovi v. Kelitu A., Exploring the biodiversity of honey- Buttel-Reepen, 1906, Apidologie 25 (1994) bees, Borneo, a Magazine on Culture, Nature, 224–238. Adventure 4 (1998) 18–33. [34] Koeniger G., Koeniger N., Tingek S., Crossfos- [47] Lahner G., Untersuchungen zur Paarungsbiolo- tered drones of Apis cerana Fabricius, 1793 and gie der Honigbiene (Apis mellifera L.) des Nyika- Apis koschevnikovi v. Buttel-Reepen, 1906 fly Massivs in Malawi, Südostafrika, Dissertation at their species specific mating times, Insectes am Fachbereich Biologie der Universität Frank- Soc. 41 (1994), 73–78. furt am Main, 1998. 338 N. Koeniger, G. Koeniger

[48] Loper G.M., Wolf W.W., Taylor O.R. Jr., Detec- [62] Plettner E., Otis G.W., Wimalaratne P.D.C., tion and monitoring of honeybee drone congre- Winston M.L., Slessor K.N., Pankiw T., Punchi- gation areas by radar, Apidologie 19 (1988) hewa R.W.K., Species- and caste-determined 103–106. mandibular gland signals in honeybees (Apis), [49] Maa T.C., An inquiry into the systematics of the J. Chem. Ecol. 23 (1997) 363–377. tribus Apidini or honeybees (), [63] Punchihewa R.W.K., Beobachtungen und Experi- Treubia 21 (1953) 525–640. mente zur Paarungsbiologie von Apis cerana [50] Mayr E., Species and Evolution, Har- indica in Sri Lanka, Dissertation am Fachbereich vard Univ. Press, Cambridge, MA, 1963. Biologie der Universität Frankfurt am Main, 1992. [51] McEvoy M., Underwood B.A., The drone and [64] Punchihewa R.W.K., Koeniger N., Koeniger G., species status of the Himalayan honey bee Apis Congregation of drones in laboriosa (Hymenoptera: Apidae), J. Kans. Ento- the canopy of trees in Sri Lanka, Apidologie 21 mol. Soc. 61 (1988) 246–249. (1990) 201–208. [52] Mogga J.B.B., Comparative studies of the indige- [65] Punchihewa R.W.K., Koeniger N., Koeniger G., nous honeybee (Apis mellifera yemenitica) and Mating behaviour of Apis cerana in Sri Lanka, the recently introduced dwarf honeybee (Apis in: Social and the Environment. Proc. florea) in Sudan, Ph.D. thesis, Dept. of Crop 11th Congr. IUSSI, 1990, p. 108. Protection, Faculty of Agriculture, Univ. Khar- toum, Sudan, 1994. [66] Rinderer T.E., Koeniger N., Tingek S., Mardan M., Koeniger G.A., Morphological comparison [53] Mogga J.B.B., Ruttner F., Apis florea in Africa: of the cavity dwelling honeybees of Borneo Apis source of the founder population, Bee World 69 koschevnikovi v. Buttel-Reepen, 1906 and Apis (1988) 100–103. cerana Fabricius, 1793, Apidologie 20 (1989) [54] Moritz R.F.A., Kryger P., Koeniger N., Estoup 405–411. A., Tingek S., High degree of polyandry in Apis [67] Rinderer T.E., Oldroyd B.P., Wongsiri S., dorsata queens detected by DNA microsatellite Sylvester H.A., De Guzman L.I., Potichot S., variability, Behav. Ecol. Sociobiol. 37 (1995) Sheppard W.S., Buchman S.L., Time of drone 357–363. flight in four honeybee species in southeastern [55] Oldroyd B.P., Smolenski A.I.J., Cornuet J.-M., Thailand, J. Apic. Res. 32 (1993) 27–33. Wongsiri S., Estoup A., Rinderer T.E., Crozier [68] Rinderer T.E., Stelzer A., Oldroyd B.P, Tingek S., R.H., Levels of polyandry and intracolonial Levels of polyandry and intracolonial relation- genetic relationship in Apis florea, Behav. Ecol. ships in Apis koschevnikovi, J. Apic. Res. 37 Sociobiol. 37 (1995) 329–335. (1998) 281–287. [56] Oldroyd B.P., Smolenski A.I.J., Cornuet J.-M., [69] Ruttner F., Die Sexualfunktion der Honigbienen Wongsiri S., Estoup A., Rinderer T.E., Crozier im Dienst ihrer sozialen Gemeinschaft, Z. Vgl. R.H., Levels of polyandry and intracolonial Physiol. 39 (1957) 577–600. genetic relationship in Apis dorsata [70] Ruttner F., The life and flight activity of drones, (Hymenoptera: Apidae), Ann. Entomol. Soc. Bee World 47 (1966) 93–100. Am. 89 (1996) 276–283. [71] Ruttner F., Drohnen von Apis cerana auf einem [57] Oldroyd B.P., Clifton S., Wongsiri S., Rinderer Drohnensammelplatz, Apidologie 4 (1973) T.E., Sylvester H.A., Crozier R.H., Polyandry 41–44. in the genus Apis, particularly Apis andreni- [72] Ruttner F., Ein metatarsaler Haftapparat bei den formis, Behav. Ecol. Sociobiol. 40 (1997) 17–26. Drohnen der Gattung Apis (Hymenoptera), Ento- [58] Oldroyd B.P., Clifton S., Parker K., Wongsiri S., mol. Ger. 2 (1975) 22–29. Rinderer T.E., Crozier R.H., Evolution of mating [73] Ruttner F., Biogeography and Taxonomy of behavior in the genus Apis and an estimate of Honeybees, Springer Verlag, Berlin, 1988. mating frequency in Apis cerana (Hymenoptera: [74] Ruttner F., Naturgeschichte der Honigbienen, Apidae), Ann. Entomol. Soc. Am. 91 (1998) Ehrenwirth Verlag, Munich, 1992. 700–709. [75] Ruttner F., Ruttner H., Untersuchungen über die [59] Pain J., Ruttner F., Les extrait de glandes Flugaktivität und das Paarungsverhalten der mandibulaires des reines d’abeilles attirent les Drohnen. II. Beobachtungen an Drohnensam- males lors du vol nuptial, C. R. Acad. Sci. Paris melplätzen, Z. Bienenforsch. 8 (1965) 1–9. 256 (1963) 512. [76] Ruttner F., Kaissling K.E., Über die interspezi- [60] Palmer K.A., Oldroyd B.P., Evolution of multi- fische Wirkung des Sexuallockstoffes von Apis ple mating in the genus Apis, Apidologie 31 mellifera und Apis cerana, Z. Vgl. Physiol. 59 (2000) 235–248. (1968) 362–270. [61] Patinawin S., Wongsiri S., Male genitalia of [77] Ruttner F., Ruttner H., Untersuchungen über die honey bees, in: Asian Apiculture. Proc. 1st Int. Flugaktivität und das Paarungsverhalten der Conf. on Asian Honey Bees and Bee Mites, Drohnen. IV. Zur Fernorientierung und Bangkok, Thailand. Wicwas Press, Cheshire, Ortsstetigkeit der Drohnen auf ihren Paarungs- CT, USA, 1993, pp. 110–116. flügen, Z. Bienenforsch. 9 (1968) 259–265. Reproductive isolation among species of the genus Apis 339

[78] Ruttner F., Ruttner H., Untersuchungen über die [94] Tingek S., Mardan M., Rinderer T.E., Koeniger Flugaktivität und das Paarungsverhalten der N., Koeniger G., Rediscovery of Apis vechti Drohnen. V. Drohnensammelplätze und (Maa, 1953): the Saban honey bee, Apidologie Paarungsdistanz, Apidologie 3 (1972) 203–232. 19 (1988) 97–102. [79] Ruttner F., Maul V., Experimental analysis of [95] Tribe G.D., Drone mating assemblies, S. Afr. reproductive interspecies isolation of Apis mel- Bee J. 54 (1982) 99–100; 103–117. lifera L. and Apis cerana Fabricius, Apidologie 14 (1983) 309–327. [96] Trjasko V.V., Über Drohnen welche sich mit Königinnen paaren, Pchelovodstvo 34 (1957) [80] Ruttner F., Woyke J., Koeniger N., Reproduction 29 – 31 (in Russian). in Apis cerana. 1. Mating behaviour, J. Apic. Res. 11 (1972) 141–146. [97] Underwood B.A., Time of drone flight of Smith, 1871 in , Apidologie 21 [81] Ruttner F., Woyke J., Koeniger N., Reproduction (1990) 501–504. in Apis cerana. 2. Reproductive organs and nat- ural insemination, J. Apic. Res. 12 (1973) 21–34. [98] Verma L.R.., Beekeeping in integrated moun- [82] Ruttner F., Kauhausen D., Koeniger N., Posi- tain development, Oxford & IBH Publ., New tion of the red honey bee, Apis koschevnikovi v. Delhi, India, 1991, p. 237. Buttel-Reepen, 1906, within the genus Apis, Api- [99] Woyke J., Natural and artificial insemination of dologie 20 (1989) 395–404. queen honeybees, Pszcz. Zesz. Nauk. 4 (1960) [83] Ruttner H., Ruttner F., Untersuchungen über die 183–275. Flugaktivität und das Paarungsverhalten der [100] Woyke J., Wovon hängt die Zahl der Spermien Drohnen. lll. Flugweite und Flugrichtung der in der Samenblase der auf natürlichem Wege Drohnen, Z. Bienenforsch. 8 (1966) 332–354. begatteten Königinnen ab?, Z. Bienenforsch. 8 [84] Seeley T.D., Honeybee Ecology, Princeton Univ. (1966) 236–248. Press, Princeton, NJ, 1985. [101] Woyke J., Natural and artificial insemination of [85] Sharma P.J., Observations on the swarming and Apis cerana in India, J. Apic. Res. 14 (1975) mating habits of the Indian honeybee, Bee World 153–159. 41 (1960) 121–125 [102] Woyke J., Rearing and instrumental insemination [86] Shearer D.A., Boch R., Morse R.A., Laigo F.M., of Apis florea queens, in: Asian Apiculture, Occurrence of 9-oxodec-trans-2-enoic acid in Proc.1st Int. Conf. on Asian Honey Bees and queens of Apis dorsata, Apis cerana and Apis Bee Mites, Bangkok, Thailand, Wicwas Press, mellifera, J. Insect Physiol. 16 (1970) Cheshire, CT, USA, 1993, pp. 104–109. 1437–1441. [103] Yoshida T., Comparative studies on the mating [87] Sheppard W.S., Honey bee evolution: molecular system of Japanese honeybee Apis cerana japon- systematic perspectives, Proc. Apimondia Congr. ica Radoszkowski and European honeybee Apis XXXVl, 1999, p. 117. mellifera L. (Hymenoptera: Apidae), Bull. Fac. [88] Simpson H., The male genitalia of Apis species, Agric. Tamagawa Univ. No. 35, 1995. Nature 185 (1960) 56. [104] Yoshida T., Yamazaki M., Difference in drone [89] Skorikov A.S., Eine neue Basis für eine Revi- congregation areas of Apis mellifera and Apis sion der Gattung Apis L., Rep. Appl. Entomol. 4 cerana japonica as a reproductive isolation (1929) 249–264 (in Russian with German sum- mechanism, in: Connor L.K., Rinderer T., mary). Sylvester A., Wongsiri S. (Eds.), Asian Apicul- [90] Strang G.E., A study of honey bee drone attrac- ture, Proc, 1st Symp. on Asian Honey Bee and tion in the mating response, J. Econ. Entomol. 63 Bee Mites, Bangkok, Thailand, Wicwas Press, (1970) 641–645, Cheshire, CT, USA, 1993, pp. 99–103. [91] Taber S., III. Factors influencing the circadian [105] Yoshida T., Saito J., Kajigaya N., The mating flight rhythm of drone honey bees, Ann. Ento- flight times of native Apis cerana japonica mol. Soc. Am. 57 (1964) 769–775. Radoszkowski and introduced Apis mellifera L. [92] Tan N.Q., Mardan M., Thai P.H., Chinh P.H., in sympatric conditions, Apidologie 25 (1994) Observations on multiple mating flights of Apis 353–360. dorsata queens, Apidologie 30 (1999) 339–346. [106] Zmarlicki C., Morse R.A., Queen mating drones [93] Thornhill R., Alcock J., The Evolution of Insect apparently congregate in certain areas to which Mating Systems, Harvard Univ. Press, Cam- queens fly to mate, Am. Bee J. 103 (1963) bridge, MA, London, 1983. 414–415.