Relationships between flowers and pollinators are generally considered cases of mutualism since both agents gain benefits. Fine-tuned adaptations are usually found in the form of strict one-to-one coevolution between species. Many insect pollinators are, however, considered generalists, visiting numerous kinds of flowers and many flower species (angiosperms) are also considered generalists, visited by many insect pollinators. We here describe a fine-tuned coevolutionary state of a flower-visiting bee that collects both nectar and pollen from an early spring flower visited by multiple pollinators. Detailed morphology of the bee proboscis is shown to be finely adjusted to the floral morphology and nectar production of the flower. Behavioral observations also confirm the precision of this mutualism. Our results suggest that a fine-tuned one-to-one coevolutionary state between a flower species and a pollinator species might be common, but frequently overlooked, in multiple flower-pollinator interactions.
Pollination is one of the commonest forms of mutualism between plants and animals1,2,3,4. Angiosperms produce flowers with nectar and ample pollen to attract flower-visitors; in return, the flower-visitors transport pollen and effect pollination, so producing seeds. These plant-pollinator interactions are predicted to be the products of coevolution between plants and insects4,5. A strict one-to-one coevolution is usually found between one plant species and its symbiotic insect partner, e.g., fig plants and fig wasps6,7. We also find fine-tuned morphological specializations in both particular flowering plants and their specialized insect pollinators8,9,10. However, many angiosperms are visited by myriad pollinators and many pollinators visit a multitude of flowering plant species. These flowers and pollinators are considered generalists5,11. The optimal phenotype of a generalist plant is expected to be the intermediate, e.g., the weighted average of all pollinators' phenotypes12. Thus we would not expect a fine-tuned specialist in mutualistic interactions between multiple flowers and multiple pollinators.
Contrary to this expectation, we here describe a fine-tuned morphological specialization between an andrenid bee (Andrena (Stenomelissa) lonicerae) and an early spring flower (Lonicera gracilipes) visited by multiple pollinators. This flower produces nectar almost exclusively for this bee. We show that the detailed functional morphology of the head and proboscis of the bee is finely adjusted to the morphology and nectar production of the flower. We also demonstrate this fine-tuned specialization from the behavioral repertoire of the bee. We then discuss the implication of this finely tuned one-to-one mutualistic state in the context of the coevolution of pollination interactions between the bee and the flowering plant.
Flower visits and nectar
We investigated the pollination activities of females of an andrenid bee, A. lonicerae, visiting the flowers of L. gracilipes (Fig. 1, Supplementary Videos S1–S3). This andrenid bee is oligolectic (collecting pollen from a limited number of phylogenetically related plants) and is found mostly on L. gracilipes13,14,15, though it occasionally visits other flowers (Table 1). Lonicera gracilipes blooms in early spring (March to May). Because very few other flowers are available in early spring, L. gracilipes is visited frequently by many insects, including another andrenid bee, Andrena hebes and halictid bees, Lasioglossum spp. (Table 2, Supplementary Fig. S1), although the commonest visitor is A. lonicerae. Pollination experiments (Supplementary Fig. S5) reveal that these other visitors may be as equally effective pollinators of this flower as A. lonicerae. However, these species collect only pollen, the nectar of this flower being almost exclusively collected by A. lonicerae (Table 2). Because the flower of L. gracilipes has a characteristic long narrow corolla tube (tube-like shape) (Supplementary Fig. S2), pollinators with a short tongue (mouthparts) have no access to its nectar (Fig. 2, Supplementary Fig. S1, Supplementary Video S5). Remarkably, even though the bee belongs to a group of short-tongued bees (Andrenidae), A. lonicerae has an extremely elongate distal part of the tongue (Figs 1d, 2a, d, m, Supplementary Fig. S3a, Supplementary Videos S3, S4). Thus, among all pollinators associated with this flower species, the principal flower-pollinator interaction appears to be that between A. lonicerae and L. gracilipes.
Females of A. lonicerae visited flowers of L. gracilipes to collect nectar (their own energy source) (Fig. 1a) and pollen (their larval food) (Fig. 1b). The foraging activities of the bee indicated that, when pollen was the first target of collection, bees almost always sought pollen before collecting nectar (Fig. 3a). In contrast, when they first took nectar, they rarely looked for pollen (Fig. 3a). There are two probable reasons for this: (1) pollen has already been removed from the flower; and (2) bees are in the pre-nesting stage (no requirement for pollen collection). This asymmetry in behavioral sequence might be explained by the difference between the two resource types: pollen and nectar.
Although a single flower has five anthers (pollen sacks), pollen is a limited non-renewable resource. Because pollen is typically removed by bees in the first couple of days of bloom, the majority of flowers in a blooming period have little pollen (Fig. 3d). When pollen is available, the bees collect it first (Fig. 3a). The limitation of pollen is also suggested by flower-breaking behavior (Fig. 1c): A. lonicerae bees sometimes tear off un-opened flower buds by their mandibles and fore legs to collect pollen, nectar or both.
In contrast to pollen, nectar is an extremely limited but renewable resource. During the brief flowering period for each blossom (ca. 4 days), the quantity of nectar secreted daily decreased rapidly from the first (0.53 ± 0.71 μl) to the sixth day (0.03 ± 0.03 μl) (Fig. 3b). Because the amount of nectar per flower is always low (0.12 ± 0.16 μl, n = 70) and many flowers have no nectar (23/70 flowers: 33%) (open flowers; Fig. 3c), the bees must visit the flowers frequently to collect nectar even if pollen is not available (850 visits out of 1171; Fig. 3a). The high frequency of flower-visits of A. lonicerae for nectar suggests that L. gracilipes provides nectar almost exclusively to this bee.
Morphology and function of the head and mouthparts
We compared the morphology of the head and mouthparts of A. lonicerae females with that of the distantly related A. (Euandrena) hebes and closely related A. (Stenomelissa) halictoides (Fig. 2, Supplementary Fig. S4, Supplementary Table S4). Sympatric and co-occurring A. hebes is the second commonest visitor to this flower, but does so only for pollen collection (Table 2). In contrast, A. halictoides, mostly allopatric to A. lonicerae, is a closely related species that utilizes the funnel-like flower of Weigela hortensis, which blooms in mid May15.
Among the three species, the female head is extremely elongate in A. lonicerae and considerably so in A. halictoides, compared with that of A. hebes (Fig. 2, Supplementary Video S5). The length/width proportion of the head is significantly different among the three species, ca. 1.10, 1.00 and 0.92 in A. lonicerae, A. halictoides and A. hebes, respectively (Supplementary Table S4). The oculo-malar space of the head is also considerably elongate in A. lonicerae and A. halictoides (c in Fig. 2a and b) compared with the more typical short malar space of A. hebes (c in Fig. 2c). The proportion of the malar space to the head width is also significantly different among species, ca. 0.12, 0.09 and 0.04 in A. lonicerae, A. halictoides and A. hebes, respectively (Supplementary Table S4). This suggests that the elongation of the head of A. lonicerae is an adaptation to tube-shaped flowers, such as those of L. gracilipes, the floral morphology of which is described below; the same applies to A. halictoides on W. hortensis15.
The mouthparts of both A. lonicerae and A. halictoides females also exhibit distinctive specializations (Fig. 2). The distal part of the tongue (glossa) is greatly extended in A. lonicerae (i in Fig. 2a; Fig. 2d, m), resembling a string and fairly elongate in A. halictoides (Fig. 2b, e), compared with that of A. hebes (Fig. 2f, n). We should note that, among approximately 1,500 species of the genus Andrena, A. lonicerae and A. halictoides are the only two species possessing an elongate distal portion of the glossa, hence their placement in the subgenus Stenomelissa. The paraglossa covering the basal part of the glossa laterally is rudimentary in both A. lonicerae and A. halictoides (j in Fig. 2d and e, b in Supplementary Fig. S4c compared with j in Fig. 2f and b in Supplementary Fig. S4b).
In both A. lonicerae and A. halictoides, there is also a distinctive modification in the maxillary galea covering the basal proboscis dorsally (f in Fig. 2a). The brush-like structure called the galeal comb (cleaning apparatus located on the inner face of the galea) shared with almost all short-tongue bees has undergone a reduction to one to three bristles (setae) (k in Fig. 2g and h, cf. k in Fig. 2i, Supplementary Table S5). The loss of almost all setae in the galeal comb and the considerable reduction in the size of the paraglossa are unique features found elsewhere in the long-tongued bees16,17, but are atypical of short-tongued bees, indicating convergence between long-tongued bees and A. lonicerae and A. halictoides.
In addition to the gross anatomy of the glossa of A. lonicerae, we examined its ultrastructure (basal part: Fig. 2j; distal parts: Fig. 2k, l). The anterior surface of the distal portion is covered with long hair-like cuticular processes (annular hairs) (l in Fig. 2l) as in the basal part of the glossa (h in Supplementary Fig. S4j), while its posterior portion has a single row of arched spine-like cuticular processes (seriate hairs) on both sides (m in Fig. 2l). The embedded cross section of this portion reveals that its posterior is a dense matrix of tissue, while the anterior contains mostly body fluid (Fig. 2l, Supplementary Fig. S4k, m), allowing flexibility. The row of cuticular processes along the lateral margins of the distal glossa of A. lonicerae and A. halictoides, which forms a tunnel-like space suitable for holding nectar, is distinctly different from the capillary (hairy) structure of the long-tongued bees, where the glossa is covered with transverse rows of slightly flattened, similar cuticular processes both anteriorly and posteriorly (Supplementary Fig. S4n–q). The elongation of the distal glossa, prementum and malar space, together with the narrow head of the bee (Fig. 2a, d, Supplementary Table S4), are thus finely tuned for extracting nectar from the base of the corolla tube (Fig. 1a, e, Supplementary Fig. S3a, Supplementary Videos S1, S3).
Molecular phylogeny of bees
We examined the molecular phylogenetic relationship among A. lonicerae, its morphological sister species A. halictoides14 and other Andrena bees, using the maximum likelihood method and Bayesian method. We confirmed that A. lonicerae and A. halictoides are monophyletic sister species distantly related to all other examined Andrena species (Fig. 5, Supplementary Fig. S6). The sequence data also indicated that A. lonicerae is derived from the ancestral A. halictoides. We also verified the current results using the neighbor joining and maximum parsimony analyses.
Floral morphology and behavioral verification
Detailed measurements of A. lonicerae reveal a precise correspondence with the tube-like flower of L. gracilipes (Fig. 4). Because nectar quantities tend to be very small, the bee has to extend its proboscis fully to collect any available nectar at the base of the funnel-shaped flower (Fig. 1e). We found extremely good agreement between the effective length (EL) of the bee's head and proboscis (see “MATERIALS AND METHODS” in Supplementary Information) and that of the corolla tube in both the Hino and Kyoto populations (Fig. 4). From the nectar-reaching rate r, all bees (100%) have an access to nectar in all flowers in the Hino population and 91% of bees in the Kyoto population. At the same time, the departure index is 0.76 mm in the Hino population and 0.00 mm in the Kyoto population. These two measures indicate the very close match between the proboscis length and corolla tube length in both the Hino and Kyoto populations.
Using video recording, we investigated the ability of the bee to reach nectar at the base of the flower (Fig. 1e, Supplementary Video S1). The bee puts its head deep into the flower (corolla tube) when it searches for nectar (Fig. 1a, e), whereas it hangs on the flower without inserting its head into the corolla when collecting pollen from anthers (Fig. 1b, Supplementary Video S2). When sucking nectar, the bee extends its mouthparts within the corolla tube (Supplementary Fig. S3a, Supplementary Videos S1, S3). The proboscis, which is folded three times under the head, is highly flexible and fully extended when necessary (Fig. 1d, Supplementary Video S4). The motion picture silhouettes of a bee sucking nectar show that the apex of the glossa reaches the base of the flower (Fig. 1e, Supplementary Video S1). These results demonstrate a fine one-to-one morphological match between the bee's proboscis and the corolla tube. Such morphological specializations, together with the behavioral data that more than 70% of flower visits are solely for obtaining nectar (Fig. 3a), indicate that A. lonicerae is mostly dependent on L. gracilipes as its source of nectar.
Thus both morphological and behavioral adaptations of the bee, as well as nectar production by the flower, indicate that this flower-pollinator interaction is highly mutualistic. This tight mutualism is strongly supported by the close matching between the effective length of the head and mouthparts of the bee and that of the corolla tube, in two separate populations (Fig. 4). The behavioral video recordings demonstrate a fine-tuned nectar collecting behavior (Fig. 1e, Supplementary Video S1).
In most pollination systems, flowering plants are typically visited by many pollinator species, while a given pollinator often visits many different flower species2,3. In such cases both flowers and pollinators are often considered generalists12,18. The bee A. lonicerae visits mostly the flower L. gracilipes and, rarely, other flowers, exhibiting narrow oligolecty14,19 (Table 1). This vernal flower otherwise appears as a generalist in that it is visited by many different insects (at least 11 species; Table 2). Hence the relationship examined here embodies an interaction between an oligolectic bee and a generalist flower. The close matching of the head and mouthparts of the bee to the tube length of the flower indicates that both the flower and the bee might have coevolved (i.e., have responded to one another evolutionarily) (Fig. 4).
Our data (Supplementary Fig. S5) show that the pollination effectiveness of one visit by other frequently visiting bees to L. gracilipes (e.g., A. hebes) is almost equivalent to that of A. lonicerae. However, assuming these bees constitute evolutionary partners (as effective pollinators) of L. gracilipes, its nectar would be expected to be accessible to these bees. The bees, however, cannot gain access to the nectar because their mouthparts are too short to reach the base of the corolla tube (Fig. 2n, Supplementary Fig. S1, Supplementary Video S5). This indicates that they are only riders, i.e., free loaders. The following preliminary observations provide support: A. lonicerae frequently displayed multiple visits from flower to flower for nectar (see also Fig. 3a), whereas the other visiting bees tended to abandon a patch of flowers after a single visit to a flower for pollen.
Close examination of the interaction between A. lonicerae and L. gracilipes reveals a tight one-to-one relationship. Among all visitors to L. gracilipes, A. lonicerae is the most frequent and essentially the only one to collect both nectar and pollen (Table 2). Although nectar production is very small in this flower (on average 0.12 μl/open flower in Fig. 3c), 72.6% of the bee's visits are solely for nectar (Fig. 3a). These data indicate that the flower is the main nectar source for the bee. Thus the present bee-flower relationship differs considerably from a generalist pollination system, even though multiple pollinators visit L. gracilipes flowers and A. lonicerae bees occasionally visit other flowers.
The phylogenetic background of A. lonicerae also supports this fine-tuned specialization between the bee and the flower. Among the approximately 1,500 species of Andrena classified as short-tongued bees, all are known to have a short tongue except four species. Among these exceptions, two species, A. (Iomelissa) violae and A. (Callandrena) micheneriana, occur in the USA17,20,21,22 and are not closely related to the other two species, A. (Stenomelissa) halictoides, which occurs in China, Korea, Japan and Russian Far East and A. (S.) lonicerae, which is endemic to Japan14,23. From morphological specializations of the head and mouthparts (malar space, glossa, paraglossa and galeal comb in Fig. 2a, b, d, e, g, h, m), the two East Asian species, belonging to the subgenus Stenomelissa, are definitely closely related sister species, distinct from other species of Andrena. The current molecular analyses support the monophyly of the two species (Fig. 5, Supplementary Fig. S6). The slightly elongate, string-like distal glossal portion of A. halictoides (Fig. 2b, e) has been further elongated in A. lonicerae (Fig. 2a, d, m). The elongation of the mouthparts of A. lonicerae is also reflected in the total length of the proboscis (Supplementary Table S4). From both morphological specialization and molecular phylogenetic data, A. lonicerae appears to be derived from the ancestral A. halictoides (Fig. 5, Supplementary Fig. S6).
The geographic distributions of both the bees and their floral hosts also support this evolutionary scenario. Most Andrena bees exhibit oligolectic (larval) pollen diets, exhibiting fairly tight relationships with their host flowers24. In Japan, nectar diet specialization is also found in the two Andrena bee species14,15: Andrena halictoides is fairly oligolectic on Weigela hortensis, but is rarely seen on many different flowers, while A. lonicerae is more (but not strictly) oligolectic on L. gracilipes. These two bee species, together with their corresponding flowering plant partners, exhibit a nearly allopatric distribution, with only narrow overlapping regions15,25. Andrenahalictoides and its host plant W. hortensis occur on the cooler Japan Sea side of Honshu and Hokkaido. Although the bee is also distributed in the eastern parts of the Asian Continent, the distributions of the bee and its host plant are almost coincident with each other in Japan. In contrast, A.lonicerae and its host L. gracilipes occur on the warmer Pacific Ocean side of Honshu and their distributions coincide nearly perfectly. The flowering period of L. gracilipes is early spring (March–May) on the Pacific Ocean side, while that of W. hortensis is late spring (May–June) on the Japan Sea side. The flowering periods of the two species may be less distinct or even overlap in northern and mountainous Honshu, where both plant species may occur in the same locations. In fact, there is a record of A. lonicerae visiting flowers of W. hortensis (Table 1). Hayashibara et al.15 found that some females of A. lonicerae visit flowers of W. hortensis to forage for pollen after the flowering period of L. gracilipes. We thus suggest that A. lonicerae originated from an ancestral population of A. halictoides in the region where W. hortensis and L. gracilipes overlapped, undergoing a shift from Weigela to Lonicera and spreading its distribution from the Japan Sea side of Honshu to the Pacific side.
Based on their molecular phylogenetic studies of Lonicera, Theis et al.26 suspected that L. caerulea is the sister species of L. gracilipes, although they included only eight of the 21 Japanese species in their analyses (note that the species name L. caelurea is misspelled as L. coerula in the tree of Fig. 1 and as L. coerulea in Appendix 1). Lonicera caerulea is widely distributed in the cool-temperate and temperate Northern Hemisphere from Europe through Japan to North America, with several named subspecies. Lonicera caerulea subsp. edulis is distributed from eastern Siberia to the Korean peninsula, as well as in Hokkaido and the subalpine zone of central and northern Honshu27. In contrast, L. gracilipes is endemic to Japan, ocurring mostly in the mountainous regions of the southeastern regions of Japan27. The two species are thus allopatrically distributed.
Judging from the wide distribution of L. caerulea subsp. edulis and the peripheral distribution of L. gracilipes in warmer climatic zones, the latter is suspected to have evolved from the former. The phenological and morphological specializations of L. gracilipes also suggest that L. gracilipes is likely to have evolved from the ancestral lineage of L. caerulea. The two flowering plants are distinct in their phenology and morphology: L. gracilipes blooms in March to April, while L. caerulea subsp. edulis blooms in May to July. The tubular part of the corolla of L. gracilipes is 10–12 mm, which is almost unique among the Japanese species of Lonicera, while that of L. caerulea subsp. edulis is 5–8 mm in (Table 3).
According to Kato28, Lonicera is a taxon having coevolved with bumblebees, although the three Japanese species, L. japonica, L. hypoglauca and L. affinis (all with very long corolla tubes) are pollinated by hawk moths. Similarly, some North American Lonicera species with elongated red corolla tubes are pollinated by hummingbirds. Kato presumed that, excepting the above three species, the effective pollinators of all Japanese Lonicera are bumblebees, typifying the insect visitors to eight Lonicera species.
From the above discussions, we suspect that L. caerulea subsp. edulis first had expanded its range to southeastern Japan, adapting to the warmer, sunnier and much less snowy winter climate of that found currently in southeastern Japan. One important aspect of this complex history is a suspected pollinator shift from bumblebees to A. lonicerae or A. halictoides. Two floral traits are characteristic of L. gracilipes: 1) the corolla tube length is much longer in L. gracilipes than in L. caerulea sub. edulis (Table 3); 2) the quantity of nectar provided by L. gracilipes is much lower than that of many flowering species typically visited by bumblebees. For example, nectar production of Weigela coraeensis, which is utilized mostly by two species of bumblebees (Bombus ardens and B. diversus) is 2.2 ± 3.0 μl/bagged flower/day (mean ± SD, n = 304; original data) in volume. In contrast, nectar production of L. gracilipes is 0.3 ± 0.51 μl/bagged flower/day (mean ± SD, n = 246). Hence ancestral L. gracilipes populations may have undergone an elongation of the corolla tube and a reduction in nectar output during their evolution from the ancestral L. caerulea stock. Meanwhile, L. gracilipes presumably underwent a shift from an unknown pollinator (bumblebees or the ancestral A. halictoides) to A. lonicerae. Elucidating this process of pollinator shifting at the origins of L. gracilipes requires identifying the pollination syndrome and current pollinators of L. caerulea. We conclude that our finding of a one-to-one matching between A.lonicerae and L. gracilipes demonstrates a high probability that coevolution has occurred between them.
Coevolution in pollination is often considered in the context of specialists and generalists, where both pollinators and flower specialists have evolved from generalists and vice versa4,5. The reason why we cannot identify the mutual partner easily is that there may be many so-called rider species (e.g., free loaders, lobbers/thieves) that utilize or steal pollen and nectar from the flower. A possible exception is social bees that collect nectar and pollen from available flowers in their surrounding environment for a long period19. As with L. gracilipes, many apparently generalist flowers may be in fact finely tuned to a single pollinator. The oligolectic pollen diet of Andrena bees also constitutes evidence for this narrow specialization24. In almost all symbiotic and mutualistic systems, we may find species-specific one-to-one interactions.
Unlike many other mutualistic (symbiotic) phenomena typically exhibiting a one-to-one relationship, pollination has been widely considered to consist of multiple interactions between many pollinators and many flowers (e.g., wild flower gardens in highlands and mast flowering in tropical rain forests4). Only a few special cases of one-to-one pollination interactions are known, e.g., fig wasps and fig trees6,7 and some flowers with long nectar spurs and their specialized pollinators (bumblebees, hawk moths and hummingbirds)9. These are exclusive interactions involving no other pollinators, nor other flowering plants. Our findings suggest a possible mutual interdependence, but not necessarily a strictly exclusive interaction. In the context of a broad coevolutionary spectrum ranging from multiple interactions to one-to-one exclusive interactions in pollination systems, we may expect many non-exclusive one-to-one interactions and their intermediate ones hidden within superficially multiple interactions. Considering the stability of pollination systems, one-to-one interactions are likely to be favorable, as opposed to the one-to-many or many-to-many interactions, typical in other mutualisms.
With respect to pollination ecology, our findings imply that co-adaptation (coevolution) and mutual relationships (interactions) are two different things. Multiple mutual interactions do not mean multiple co-adaptations. Many angiosperms may have a target pollinator that guarantees the pollination of their own flowers. Many insect pollinators with only a short active period may also have a target plant species when they search for flowers, as does A. lonicerae. Hence one-to-one co-adaptation might have evolved and been retained within multiple pollination interactions.
We here describe a fine-tuned morphological specialization between an andrenid bee (Andrena (Stenomelissa) lonicerae) and an early spring flower (Lonicera gracilipes) visited by multiple pollinators. This flower produces nectar almost exclusively for this bee. We show that the detailed functional morphology of the head and proboscis of the bee is finely adjusted to the floral morphology and nectar production of the flower. We also demonstrate this fine-tuned specialization from the behavioral repertoire of the bee. See Supplementary Information for the detailed methods.
Stebbins, G. L. Adaptive radiation of reproductive characteristics in Angiosperms. I: Pollination mechanisms. Annu. Rev. Ecol. Syst. 1, 307–326 (1970).
Faegri, K. & van der Pijl, L. The Principles of Pollination Ecology, 3rd edn (Pergamon Press, Oxford, 1979).
Proctor, M., Yeo, P. & Lack, A. The Natural History of Pollination (Timber Press, Portland, 1996).
Willmer, P. Pollination and Floral Ecology (Princeton Univ. Press, Princeton, 2011).
Waser, N. M. & Ollerton, J. eds Plant-Pollinator Interactions: from Specialization to Generalization (Univ. of Chicago Press, Chicago, 2006).
Cook, J. M. & Rasplus, J.-Y. Mutualists with attitude: coevolving fig wasps and figs. Trends Ecol. Evol. 18, 241–248 (2003).
Cruaud, A. et al. An extreme case of plant-insect codiversification: figs and fig-pollinating wasps. Syst. Biol. 61, 1029–1047 (2012).
Heinrich, B. Resource partitioning among some eusocial insects: bumblebees. Ecology 57, 874–889 (1976).
Whittall, J. B. & Hodges, S. A. Pollinator shifts drive increasingly long nectar spurs in columbine flowers. Nature 447, 706–712 (2007).
Anderson, B. & Johnson, S. D. The geographical mosaic of coevolution in a plant-pollinator mutualism. Evolution 62, 220–225 (2007).
Ollerton, J. Reconciling ecological processes with phylogenetic patterns: the apparent paradox of plant-pollinator systems. J. Ecol. 84, 767–769 (1996).
Aigner, P. A. Optimality modeling and fitness trade-offs: when should plants become pollinator specialists? Oikos 95, 177–184 (2001).
Hirashima, Y. Systematic and biological studies of the family Andrenidae of Japan (Hymenoptera, Apoidea). Part 2. Systematics, 6. J. Fac. Agr. Kyushu Univ. 13, 493–517 (1965).
Tadauchi, O. & Hirashima, Y. Synopsis of Andrena (Stenomelissa) with a new species from Japan (Hymenoptera, Andrenidae). J. Fac. Agri. Kyushu Univ. 33, 67–76 (1988).
Hayashibara, K., Maeta, Y. & Kitamura, K. Studies on the pollination syndrome in sympatrically distributed two caprifoliaceous plant species, Loniceragracilipes and Weigerahortensis in Sanin District, western Japan. Bull. Hoshizaki Green Found. 14, 1–30 (2011) (In Japanese with English summary).
Alexander, B. A. & Michener, C. D. Phylogenetic studies of the families of short-tongued bees. Univ. Kansas Sci. Bull. 55, 377–424 (1995).
Michener, C. D. The Bees of the World. 2nd edn (Johns Hopkins Univ. Press, Baltimore, 2007).
Waser, N. M., Chittka, L., Price, M. V., Williams, N. M. & Ollerton, J. Generalization in pollination systems and why it matters. Ecology 77, 1043–1060 (1996).
Cane, J. H. & Sipes, S. in Plant-Pollinator Interactions: from Specialization to Generalization (eds Waser, N. M. & Ollerton, J.) 99–122 (Univ. of Chicago Press, 2006).
LaBerge, W. E. A revision of the bees of the genus Andrena of the western hemisphere, Part I. Callandrena. Bull. Univ. Neb. St. Mus. 7, 1–316 (1967).
LaBerge, W. E. A revision of the bees of the genus Andrena of the western hemisphere, Part XI. Minor subgenera and subgeneric key. Trans. Am. Ent. Soc. 111, 441–567 (1986).
Laroca, S., Michener, C. D. & Hofmeister, R. M. Long mouthparts among “short-tongued” bees and the fine structure of the labium in Niltonia (Hymenoptera, Colletidae). J. Kans. Ent. Soc. 62, 400–410 (1989).
Xu, H. & Tadauchi, O. The subgenera Margandrena, Leimelissa, Stenomelissa and Zonandrena of the genus Andrena of Eastern Asia (Hymenoptera, Andrenidae). J. Fac. Agr. Kyushu Univ. 53, 63–66 (2008).
Larkin, L. L., Neff, J. L. & Simpson, B. B. The evolution of a pollen diet: Host choice and diet breadth of Andrena bees (Hymenoptera: Andrenidae). Apidologie 39, 133–145 (2008).
Takahashi, H. [Distributional records of the subgenus Stenomelissa of the genus Andrena (Hymenoptera, Apoidea).]. Tsunekibachi 16, 71–79 (2009) (In Japanese).
Theis, N., Donoghue, M. J. & Li, J. Phylogenetics of the Caprifolieae and Lonicera (Dipsacales) based on nuclear and chloroplast DNA sequences. Syst. Bot. 33, 776–783 (2008).
Hara, H. A revision of Caprifoliaceae of Japan with reference to allied plants in other districts and the Adoxaceae. Ginkgoana (5), i–iv, 1–336, pls. 1–55 (1983).
Kato, M. Coevolution system between angiosperm floras and bumblebees. 661 Shu-seibutsugaku-kenkyu. 11, 1–13 (1987) (In Japanese).
We thank H. Takahashi for identification of Andrena species and many helpful suggestions in this study, Y. Maeta and A.C. Harris for critical comments on our manuscript, H. Kurushima for help in our field investigation and X. Fu for providing three photographs (Fig. 1a–c). This work was supported by grants-in-aid from the Ministry of Education, Culture, Sports, Science and Technology of Japan to J.Y. (nos.22370010 and 22255004) and A.S. (nos. 21570100 and 25440217).
The authors declare no competing financial interests.
Electronic supplementary material
Supplementary Video S1.
Supplementary Video S2.
Supplementary Video S3.
Supplementary Video S4.
Supplementary Video S5.
Supplementary Video S6.
About this article
Cite this article
Shimizu, A., Dohzono, I., Nakaji, M. et al. Fine-tuned Bee-Flower Coevolutionary State Hidden within Multiple Pollination Interactions. Sci Rep 4, 3988 (2014). https://doi.org/10.1038/srep03988
This article is cited by
Staminode of Jacaranda rugosa A.H. Gentry (Bignoniaceae) promotes functional specialization by ensuring signaling and mechanical fit to medium-sized bees
Organisms Diversity & Evolution (2022)
Bulletin of Mathematical Biology (2019)
Comparative transcriptome analysis of oil palm flowers reveals an EAR-motif-containing R2R3-MYB that modulates phenylpropene biosynthesis
BMC Plant Biology (2017)
Journal of Insect Behavior (2016)