Pollinator and floral odor specificity among four synchronopatric species of Ceropegia (Apocynaceae) suggests ethological isolation that prevents reproductive interference
Fly pollination in African and in Asian Ceropegia
The four Ceropegia we investigated were pollinated by Diptera of two families, Chloropidae and Milichiidae, the latter represented by two genera, Milichiella and Neophyllomyza. Pollinator assemblages were significantly different between all pairs of plant species. With regard to the pollinator specialization, there are two aspects: functional specialization, i.e. interacting with a functional group of pollinators, and ecological specialization, i.e. interacting with relatively few species of pollinators. We consider the four Ceropegia species in our study as equally functionally specialized since they are pollinated by Chloropidae and Milichiidae flies, both of which are long-tongued flies and often kleptoparasites. Therefore, the way these flies interact with Ceropegia flowers may exert similar selective pressures on floral traits. Functional specialization on pollination by Diptera of only one or two families has been generally documented in Ceropegia21,22. However, ecological specialization18,19 is so far infrequent in the genus. It has been shown by Heiduk et al.24 for other species of Ceropegia from East Africa and South Africa that they are not ecologically highly specialized, because the functional group of pollinators often included several fly species. A different trend is revealed here: three of the four are ecologically specialized on pollination by only two, species-specific, pollinating fly morphospecies. Ceropegia citrina, which only relied on two fly morphospecies of the same genus, i.e., Neophyllomyza, is the most specialized. In contrast, C. boonjarasii is most generalist, being pollinated by six Chloropidae morphospecies. Each of the other two plant species, C. acicularis and C. tenuicaulis, was pollinated by two fly morphospecies of different families. In terms of the diversity of pollinator species, these two were thus more ecologically specialized than C. boojarasii, but less than C. citrina. Both Chloropidae and Milichiidae were also reported as pollinators of C. thaithongiae, another Thai endemic species33. Neophyllomyza flies were found to be pollinators of C. dolichophylla in China24,26 and visitors of C. nilotica in southern Africa21, while Milichiella flies were visitors of C. arabica var. powysii in East Africa21 and C. sandersonii in South Africa23.Large-scale studies of Ceropegia (including the stapeliads s. str.) covering the five major centers of diversity21,22, i.e., the Indian subcontinent (barring Southeast Asia), the Arabian Peninsula, East Africa, southern Africa, and West Africa, revealed flies of 16 families as pollinators. These studies also showed that the proportional use of different fly families as pollinators did not differ across all centers of diversity, thus no difference was found in this respect between Ceropegia from Africa and Asia. Moreover, in all regions, Chloropidae and Milichiidae were important pollinators. Note that the pollinator families used by Ceropegia s. str. were very different from those of the stapeliads s. str.Pollinator attraction and pollination strategyA main driving force behind floral visitation by adult Diptera, which can subsequently lead to successful pollination, is their requirement for sugar as a primary energy source, which can be obtained from floral nectar, and/or nutrients necessary for reproduction, which can be obtained from pollen35. Ceropegia flowers, however, do not offer pollen that can be consumed by flies, and are generally known to be nectarless. It has been suggested that they attract fly pollinators by deception through food source or brood site mimicry20,21,24. Nectar has thus far only been shown to be present in an African species (C. ampliata25) and a species endemic to Thailand, C. thaithongiae33. Also, a preliminary study on C. acicularis and C. tenuicaulis revealed the presence of viscous nectar in very small volume. In both species, nectar contained about 60% of fructose and 40% of glucose (unpublished data). Despite the presence of small quantities of nectar, these species can still be regarded as deceptive since they advertise another reward via scent mimicry21,22,24. As Ceropegia flowers trap flies for several hours or days, the function of nectar has been thought to be to maintain pollinators alive until they are released22,29,36. In the four Ceropegia species studied here, oviposition has not been observed; therefore, brood-site deception is unlikely.Recently, prey mimicry has been described in C. sandersonii23 and C. dolichophylla26. Both plant species were experimentally shown to be kleptomyiophilous, targeting their kleptoparasitic Desmometopa (Milichiidae) pollinators through the emission of VOCs that mimicked a particular food source for female flies. Many species of Choropidae and Milichiidae are kleptoparasites, attracted by various types of prey, frequently true bugs, killed by other arthropods. There is increasing evidence that flowers pollinated by such flies are kleptomyiophilous22,23,24,28. Typically, only females are kleptoparasitic26,37, because they require protein for egg maturation. The overwhelming majority (99%) of Diptera found in flowers of the four studied Ceropegia species were female, which is in accordance with previous findings for floral visitors of this plant group21,22. Forty percent of all major compounds identified in the four Ceropegia species are known to be released from insects, in particular by certain Hemiptera (Coreidae, Miridae, Pentatomidae). For example, (Z)-3-octen-1-yl acetate, (E)-2-octen-1-yl acetate and butyl butyrate were reported from Miridae38,39, and (E)-2-decen-1-yl acetate from Pentatomidae40. Hexyl butyrate, which was detected in the floral scents of C. citrina in a large proportion, is known to be released in prey defense secretions and responsible for the attraction of kleptoparasitic Neophyllomyza flies39,41,42,43 Hexyl acetate and 1-hexanol, the main compounds in the floral scents respectively emitted by C. tenuicaulis and C. acicularis, are known as components of the secretion of Coreidae bugs44,45. Interestingly, a species of Milichiella, the genus of flies that pollinated these two Ceropegia species, was reported to be attracted to various crushed Coreidae and Pentatomidae46. Certain VOCs have also been reported in other Ceropegia species, e.g., (E)-2-hexenyl acetate in C. sandersonii23, which is pollinated by kleptoparasitic Desmometopa flies and also visited by Milichiella flies. Aristolochia rotunda (Aristolochiaceae)28, a species with trap flowers similar to those of Ceropegia, emits volatiles similar to those of freshly killed Miridae bugs that were shown to attract its pollinating kleptoparasitic Chloropidae flies. These components included, inter alia, hexyl butyrate and octyl butyrate, which were present in C. acicularis, whose main pollinators we have also shown to be Chloropidae flies. Pollination by kleptoparasitic flies and emission of compounds also released in insect secretions support our assumption that, like other Ceropegia spp., C. acicularis, C. boonjarasii, C. citrina and C. tenuicaulis are kleptomyiophilous, and it is likely that VOCs play a major role in attracting their dipteran pollinators from a distance. In our study, a clear distinction in both the chemical profiles of VOC emissions of the four Ceropegia species and the identities of pollinating flies attracted to them suggests that each of them, if truly kleptomyiophilous, mimicks a different model. Further investigation should determine which compounds actually attract the particular pollinating flies, and which insects (models) are potentially mimicked by each species.In addition to VOCs, which appear to play a key functional role, morphological attributes, such as shape and size of floral parts, colors and patterns or other functional traits of flowers, which were very distinctive between the investigated Ceropegia species, may also affect the attractiveness of flowers. The experiments performed here on C. boonjarasii support the hypothesis that the vibratile trichomes on the corolla lobe tips play a role in attracting the fly pollinators of this species. In their natural habitats, Ceropegia generally grow close to the ground, hidden among the tall grasses or in small bushes47. It has therefore been suggested that olfactory cues are more important than visual cues in pollinator attraction20. However, in the case of the twiner C. boonjarasii, which climbs higher up in the vegetation and can present its flowers at a level up to 200 cm above the ground, the visually attractive vibratile trichomes of its flowers appear well adapted to its growth habit and the place where it grows, playing a key role in short-distance attraction, as proposed by Vogel20. Similar motile floral appendages48 are known in other Asclepiadoideae, including Stapelia spp. (Ceropegieae)49, and in plants of unrelated angiosperm families, e.g., Trichosalpinx spp. (Orchidaceae, Pleurothallidinae)50, pollinated by Diptera28,31,50.Our video recordings of flies trapped inside the flowers of C. tenuicaulis (Video 2) and C. acicularis (Video 3) are the first such recordings for Ceropegia trap flowers and showed that flies actively moved downwards to the base of the corolla inflation. This has been described by Vogel20 as positive phototaxis due to lighter-colored epidermal cells (“light windows”) around the gynostegium that direct the flies to move into contact with the reproductive organs, where they then remove pollinaria (the corpusculum being attached to the insect’s proboscis) or deposit pollinia (the guide rails take up the insertion crest of a pollinium). This uniform placement of pollinaria on different regions of insect mouthparts (e.g. rostrum, labellum or trichomes on lip pads) has been shown to be a phylogenetically conserved trait in the entire tribe Ceropegieae, which use flies, wasps, or beetles as pollinators20,21,22,29,49. In this regard, nectar (known to be present in certain species, see above) or probably minimal amounts of sugar-containing secretion contained in the cups formed by the corona appendages around the central gynostegium, may play a role in positioning the pollinators correctly.Entrapment of flies by the tubular corolla and guidance of trapped flies to the floral reproductive organs by the light windows are the important mechanisms ensuring successful pollination in Ceropegia. The circumstance observed on C. acicularis—most of whose Chloropidae pollinators moved purposefully downward into the corolla tube, but often left the flower after a short while—implies that the flower of this species is less efficient in trapping flies. Interestingly, the main pollinators of this species preferentially visited its flowers at the end of the day, when the light intensity had decreased drastically. Therefore, once flies entered the flower, whose opening at the top of the corolla tube (i.e., the fly’s exit) was already obscured by the darkness outside, their directional movement in response to light was impeded, forcing flies to stay overnight inside the flower. Whether these circumstances reflect coincidence or alternatively specialization for pollination by flies that seek overnight shelter, they seem to mitigate the effect of inefficient entrapment of flies by flowers of C. acicularis.Reproductive isolationThe four species of Ceropegia under consideration were found in the same habitat with individuals of different species occurring close to each other, and had largely overlapping flowering phenology. All species seem to share the same pollination strategy and attracted flies of only one or two families (Chloropidae and Milichiidae) with similar ecological traits. Their flowers were receptive all day long, and visits of fly pollinators, although varying according to the Ceropegia species they visited, largely overlapped. Overall, the combination of these factors might lead to pollinator sharing and, in turn, to interspecific pollinia transfer. Hereof, Meve and Liede-Schumann51 suggested that hybridization mediated by small flies was most likely responsible for morphological transitions between the taxa of Ceropegia s. str. and those of Brachystelma s. str. (recently merged into the former). Moreover, one of the authors (M. Kidyoo, unpublished data) found putative hybrids in various Ceropegia populations across Thailand. Thus, natural hybridization in the genus is known from co-occurring species. Nevertheless, in continued observations from 2014 to 2019, no putative hybrid plant individuals (i.e., with vegetative and floral morphology intermediate between the four studied species) have ever been found in Pha Taem National Park, indicating that species integrity is well maintained through efficient reproductive isolation among the four species. Whether or not post-zygotic reproductive isolation occurs, we expect pre-zygotic isolation mechanisms to have evolved because of the relatively high value of each pollinarium (only five per flower and few flowers per plant).The four species have diverged over time in particular floral traits. There are sharp floral morphological differences in shape, color patterns and other functional traits such as vibratile trichomes, which were shown to contribute to pollinator attraction in C. boonjarasii. Some similarities between the VOCs emitted by flowers of the four species suggest that the identities and proportions of compounds have changed over time due to selection exerted by different flies at this locality, which are all likely to be kleptoparasites but with different preferences for certain volatiles. Overall, specific pollinator attraction mediated by olfactory and visual floral cues—‘ethological isolation’—is a crucial pre-zygotic isolation mechanism to avoid interspecific visitation and pollinator limitation due to sharing. A contrasting case has been found in two other Ceropegia species, each native to different geographical regions, C. sandersonii from South Africa and C. dolichophylla from China. In that case, there is thus no selection pressure for divergence. In their natural habitats, each species uses different floral scent profiles to attract fly pollinators of the same genus, i.e., the kleptoparasitic Desmometopa spp. It has thus been proposed that these two Ceropegia exploit different olfactory preferences of flies of the same genus23,24,26.Pre-zygotic isolation can also be achieved through ‘mechanical isolation’17, requiring morphological fit either between flower and pollinator and/or between male (pollinium, in particular insertion crest) and female (guide rails) reproductive organs, known as a lock-and-key mechanism52, which can be achieved through various morphological adaptations. The role of floral proportions in regulating the fly sizes that can function as pollinators has been suggested in Ceropegia taxa21. It has also been highlighted in a stem-succulent, open-flowered stapeliad, Orbea lutea subsp. lutea, the head width of whose effective pollinating Atherigona (Muscidae) flies is small enough to fit and probe the cavity formed by the inner corona lobes beneath the guide rails for nectar53. In the present study, the size of the floral entrance, i.e. the smallest diameter of the corolla tube, was thought to be the first filter screening out visitors of the four Ceropegia flowers. However, no effect of corolla tube diameter on the fly sizes was detected. Each fly morphospecies that pollinates one of these four Ceropegia species was small enough to enter the tubular corolla of the other co-occurring species. Moreover, the pollinator exchange experiment performed in this study demonstrated that, in spite of distinctive corona structure (shape, size and spatial arrangement of corona lobes) between the investigated Ceropegia species, which concealed to varying extents the central floral reproductive organs, all fly morphospecies tested, including flies that were not natural pollinators (even a laboratory strain of D. melanogaster, a widespread Diptera of a family completely different from the identified pollinators), were able to remove the pollinaria, even those of C. citrina that appeared to be well hidden (Fig. 2G–I). Therefore, corona structure seems not to restrict visitors from removing the pollinaria. The fact that some removed pollinaria fell off and were found free inside the flowers shows that the pollinaria were not firmly attached to the experimental non-natural visitors. It is thus likely that their micromorphology did not fit properly that of the plant’s reproductive organs. Therefore, non-specific flies can lose pollinaria in flowers and are unlikely to be able to actually insert pollinia. Whatever the fate of the removed pollinia, they are lost from the flower if they do not reach another flower of the same species. The fact that pollinaria can be removed by flies that are not naturally attracted to and visiting flowers suggests that there is selective pressure for efficient ethological isolation. The distinct floral scents among Ceropegia species evidenced in this study might be the result of such selective pressure.
Another aspect of mechanical isolation is the lock-and-key relation between the guide rails (lock) and pollinium (key) that has been reported from several Asclepiadoideae, e.g., Cynanchum s. l. (former Sarcostemma)54 and the stapeliads32. Other studies, however, confirmed the lack of such a precise mechanism, e.g., in Asclepias55 and Vincetoxicum s. l. (former Tylophora)52. Here, the preliminary study of morphology of pollinaria showed strong differences between the studied Ceropegia species in the shape and size of corpuscula, translator arms and pollinia, especially the insertion crest, which is the portion that is inserted as a “key” between the guide rails (the “lock”). This result is suggestive of a possible lock-and-key mechanism.Although this first study has not yet reached a full conclusion about the mechanisms enforcing the reproductive boundaries between the four Ceropegia species, it advances our understanding of how these investigated species maintain their integrity, and highlights the crucial role of ethological isolation. Other core elements of work need to be done to exactly examine the risk of interspecific pollinia transfers and hybridization. On the one hand, the morphological fit between the guide rails and pollinia has to be inspected by a detailed micromorphological study. On the other, further field experiments must be conducted to test if a pollinator of a given plant species, e.g., C. acicularis, can insert pollinia of this plant into the guide rails of any of the other three ‘wrong’ Ceropegia species (and all possible combinations thereof). A simpler experiment would test if those fly species that are attracted to more than one species (e.g., Milichiella sp. 1) are actually capable of exporting pollen to flowers of the wrong species. More
