Lying to Pinocchio: floral deception in an orchid pollinated by long-proboscid flies

Transcription

1 Blackwell Publishing LtdOxford, UKBOJBotanical Journal of the Linnean Society The Linnean Society of London, 2006? ? Original Article LYING TO PINOCCHIO S. D. JOHNSON and S. MORITA Botanical Journal of the Linnean Society, 2006, 152, With 3 figures Lying to Pinocchio: floral deception in an orchid pollinated by long-proboscid flies STEVEN D. JOHNSON 1 * and SHELAH MORITA 2 1 School of Biological and Conservation Sciences, University of KwaZulu-Natal, P. Bag X01, Scottsville, Pietermaritzburg 3209, South Africa 2 Section of Evolution and Ecology, University of California Davis, CA 95616, USA Received July 2005; accepted for publication May 2006 Plants that lack floral rewards may nevertheless attract pollinators if their flowers sufficiently resemble those of rewarding plants. Flowers of the South African terrestrial orchid Disa nervosa are similar in floral dimensions and spectral reflectance to those of a sympatric nectar-producing irid (Watsonia densiflora s.l.). Observations showed that the orchid and Watsonia share the same pollinator, a long-proboscid tabanid fly Philoliche aethiopica. These flies visited inflorescences of both species during their foraging bouts and most (64%) observed or captured on Watsonia inflorescences carried pollinaria of the orchid on their proboscides. They probe an average of 6.3 flowers on Watsonia inflorescences, but just 1.9 flowers on the Disa inflorescences, a behaviour which would strongly promote crosspollination in the self-compatible orchid. The orchid generally achieves high levels of pollination success, with approximately 50% of flowers receiving or exporting pollen at some sites. Pollination success was also high at one site that lacked Watsonia plants, suggesting that the orchid does not have an obligate dependence on Watsonia. Its pollination system may therefore be characterized as intermediate between generalized food deception and specific floral mimicry.. ADDITIONAL KEYWORDS: adaptation Iridaceae mimicry mutualism Orchidaceae Tabanidae. INTRODUCTION Many plants do not offer floral rewards and, instead, employ various ruses to attract pollinators. This strategy is particularly prevalent in orchids (Dafni, 1987; Jersakova, Johnson & Kindlmann, 2006). Imitation of the flowers of sympatric rewarding plants is one mechanism that can allow nonrewarding plants to achieve relatively high rates of visitation by pollinators (Dafni & Ivri, 1981; Nilsson, 1983; Johnson, 1994). This form of mimicry (floral Batesian) is well-developed in some orchid genera, notably Disa Berg. (Johnson, 1994, 2000; Johnson, Alexandersson & Linder, 2003a; Anderson, Johnson & Carbutt, 2005; Anderson & Johnson, 2006). Recent field evidence suggests that even orchids with flowers that are not particularly similar in morphology to those of sympatric rewarding plants, may attract insects conditioned to associate a *Corresponding author. johnsonsd@ukzn.ac.za particular colour with a reward (Gumbert & Kunze, 2001; Kunze & Gumbert, 2001). Experimental work with caged insects has confirmed that prior colour conditioning is a key factor in attraction of pollinators to nonrewarding flowers (Gumbert, 2000; Gigord et al., 2002). Therefore, the traditional dichotomy between generalized food deception and Batesian mimicry may be an oversimplification. Non-rewarding species could, in theory, occupy any point along a continuum from vague to precise imitation of rewarding flowers (Gumbert & Kunze, 2001; Johnson et al., 2003b). Anderson et al. (2005) recently suggested that the degree of specialization in the pollination system that is being exploited will have a major influence on the degree of similarity between mimics and their models. For example, many orchids in Europe require their flowers to have only a vague resemblance to those of sympatric rewarding plants because their bumblebee pollinators, particularly newly emerged queens and workers, are highly generalized in their foraging behaviour. On the other hand, many nonrewarding 271

2 272 S. D. JOHNSON and S. MORITA orchids in South Africa are pollinated by long-proboscid flies that form specialized and predictable associations with rewarding plants (Goldblatt & Manning, 2000). In order to elicit visits from these flies, nonrewarding plants appear to require accurate matching of the flower colour and inflorescence shape of plants that these flies depend on for nectar (Johnson et al., 2003a). The South African orchid Disa nervosa Lindl. has large pink flowers with a superficial resemblance to those of a nectar-producing irid Watsonia densiflora Baker (Fig. 1A). This resemblance, together with the observation that D. nervosa is both nonrewarding and most often found intermingled with plants of W. densiflora, led us to hypothesize that the orchid is a mimic of the irid, or at least in some way exploits a specialized relationship between the irid and its pollinators. The aims of this study were to test the mimicry hypothesis by establishing whether the pollinators of D. nervosa depend on W. densiflora as their main source of nectar at sites where the orchid occurs, to compare the morphological and colour traits of the two plant species, to observe whether flies visiting Watsonia would also visit Disa plants during their foraging bouts, and to determine the effects of the presence of the putative model on pollination success of the orchid. MATERIAL AND METHODS THE STUDY SPECIES Disa nervosa occurs in open grassland habitats in the coastal zone of KwaZulu-Natal and the Pondoland region of the eastern Cape, as well as the mountains along the Swaziland South Africa border (Linder, 1981; Fig. 2). Flowering takes place mainly in February. Its putative model Watsonia densiflora is a very common species with a similar distribution to D. nervosa along the coastal zone (Fig. 2). It is replaced by Watsonia pulchra N. E. Br. ex Goldblatt in the mountains along the Swaziland South Africa border (Fig. 2). These two Watsonia species differ only in minor aspects of their bract morphology, but have otherwise virtually indistinguishable flowers. Some taxonomists have included W. pulchra within a broadly circumscribed W. densiflora (cf. Verdoorn, 1959), but W. pulchra is treated as a distinct taxon in the most recent revision (Goldblatt, 1989). Both have their peak flowering in February also. No previous studies have been conducted on the pollination of either D. nervosa or the W. densiflora complex. STUDY SITES Fieldwork for this study was conducted at seven sites in South Africa in February and March (Fig. 2; Table 1). At six of these sites, plants of D. pulchra were found intermingled with plants of Watsonia densiflora or W. pulchra (in the case of the populations at the Saddleback and Josefsdal sites on the Swaziland border). Voucher specimens from the Highover and Umtamvuna sites are deposited in the Bews Herbarium (NU) at the University of KwaZulu- Natal, Pietermaritzburg. POLLINATOR OBSERVATIONS In order to determine the pollinators of the study species, we carried out approximately 60 h of observations over 14 days (typically between and hours). The observations took place at Highover (1/24/25.ii.2005), Roselands (29.ii.2003; two other dates not recorded), Umtamvuna (30.iii.2004; iii.2005), Weza (28.ii.2004), Josefsdal and Saddleback (9.iii.2004), and Durban (ii.2003, exact dates not recorded). Insects visiting the orchids or Watsonia flowers were captured and examined for the presence of pollinaria. We also recorded pollinaria on insects that were seen foraging, but not captured. Pollinaria on the insects were compared with those of D. nervosa under a dissecting microscope. We could be confident Table 1. Description of the study sites Site Coordinates Elevation (m a.s.l.) Estimated number Disa plants Watsonia plants Highover farm S, E Roselands S, E Umtamvuna S, E Durban S, E Weza S, E Josefsdal S, E Saddleback Pass S, E

3 LYING TO PINOCCHIO 273 Figure 1. A. Grassland habitat at the Highover study site with flowering plants of Disa nervosa (left foreground) and Watsonia densiflora (right). B. Long-proboscid fly Philoliche aethiopica probing a flower of W. densiflora. Pollinaria of D. nervosa are attached to its proboscis. Scale bar = 10 mm. C. Inflorescences of D. nervosa (left) and W. densiflora (right). Scale bar = 50 mm. D. Philoliche aethiopica probing a flower of D. nervosa. Scale bar = 10 mm. E. Fly (P. aethiopica) posed next to a flower of D. nervosa. Scale bar = 10 mm. F. Philoliche aethiopica with pollinaria of D. nervosa probing a damaged flower of W. densiflora. Scale bar = 10 mm.

4 274 S. D. JOHNSON and S. MORITA The degree of similarity in the morphology of D. nervosa and W. densiflora was established by measuring flower tube length, inflorescence height, and the number of open flowers per inflorescence for plants of each species at the Highover site. Spectral reflectance of flowers of each species was measured over the UVvisible range ( nm) using the apparatus described by Johnson et al. (2003a). Nectar was not present in flowers of D. nervosa, as verified by examination of spurs of freshly picked unpollinated flowers under a dissecting microscope with backlighting, which would reveal the presence of any nectar droplets. The standing crop of nectar in 21 flowers from different plants of W. densiflora in the Highover population was measured at hours using 5-µL microcapillaries and the concentration of this nectar was determined with a 0 50% refractometer. Figure 2. Distribution of Disa nervosa and its two putative models, Watsonia densiflora and W. pulchra, based on herbarium records. Sites mentioned in the text are indicated on the map as follows: Durban (D), Highover (H), Josefsdal (J), Roselands (R), Saddleback (S), Umtamvuna (U) and Weza (W). that pollinaria on the insects originated from D. nervosa as no other orchids with sectile pollinia were in flower at these sites. Granular pollen on the bodies of the insects was removed with fuchsin gel (Beattie, 1971) and examined under a compound microscope. Pollen was identified from a reference collection taken at the study sites. Insect voucher specimens are deposited in the collections at the Natal Museum, Pietermaritzburg and the Bohart Museum, Davis, California (accession numbers 26001, and 26003). The behaviour of insects was also observed to determine (1) whether plants of W. densiflora and D. nervosa are visited in the same foraging bouts and (2) the number of flowers probed by insects during single visits to inflorescences of the two species. We used a presentation stick (Thomson, 1988) to determine whether flies foraging on a Watsonia inflorescence would choose to visit a inflorescence of D. nervosa that was proffered at a distance of one metre or fly to a nearby Watsonia inflorescence. PLANT TRAITS EFFECT OF WATSONIA ON DISA POLLINATION SUCCESS The proportion of D. nervosa flowers that received or exported pollen was estimated for plants at five of the study sites. The mean proportion of D. nervosa flowers that received or exported pollen and the distance to the nearest Watsonia plant was recorded for plants at each site. We looked at all open or recently wilted flowers on each plant at all of the sites except at Highover, where we randomly selected six flowers per plant for examination. Using a 10 hand lens, we established whether pollen massulae were present on the stigma and whether one or both of the pollinaria had been removed. The relationship between the average pollination success per plant (proportion of flowers with either pollen deposited or removed) and proximity of W. densiflora inflorescences was determined using linear regression and ANCOVA, with site as a factor and distance to W. densiflora as a covariate. RESULTS POLLINATOR OBSERVATIONS The long-proboscid fly Philoliche aethiopica Thunberg (Diptera: Tabanidae) was the only insect species found to carry pollinaria of Disa nervosa at the study sites; it was also the only insect species observed to visit flowers of this species. Of the 47 P. aethiopica flies observed on Watsonia inflorescences at the Highover site, 30 (64%) carried pollinaria of D. nervosa. Pollinaria were attached to the underside of the basal section of the proboscis, which had an average length of 15.4 mm (SD = 0.8 mm, N = 13). An average ± SD of 2.7 ± 1.7 pollinaria were recorded on 15 flies that were captured or approached closely enough to count the pollinaria. Other flies were observed to carry large pollinaria loads that could not be counted precisely and are therefore not included in this sample. Flies observed visiting flowers of D. nervosa were seen to extract pollinaria and also probe deeply enough for pollen to be deposited onto the stigma. Watsonia densiflora, like D. nervosa, appears to be specialized for pollination by P. aethiopica and is prob-

5 LYING TO PINOCCHIO 275 Table 2. Comparison of fly behaviour, morphology, and nectar properties of inflorescences of Watsonia densiflora and Disa nervosa [mean ± SD (N)] Behaviour/trait Watsonia Disa t P Flowers probed by flies 6.3 ± 3.8 (33) 1.9 ± 0.7 (12) Floral tube (mm) 19.4 ± 2.2 (31) 20.9 ± 2.4 (53) Open flowers 10.4 ± 4.6 (11) 8.3 ± 3.6 (23) Inflorescence height (cm) ± 12.1 (13) 66.3 ± 7.9 (9) 8.11 < Nectar volume (µl) 1.0 ± 0.7 (21) n/a Nectar concentration (%) 21.7 ± 0.7 (21) n/a ably the major source of nectar for this fly in the summer-rainfall coastal belt region of South Africa. These flies were frequent visitors to flowers of W. densiflora. The only other insects seen to visit flowers of W. densiflora were pollen-collecting anthophorid bees, but these were much less numerous than the flies. The flies crawl into the expanded portion of the perianth tube, beneath the anthers and stigma, and insert their proboscis into a narrow, constricted portion that contains c. 1.0 µl of nectar with a concentration of c. 22% (Table 2). Individuals of P. aethiopica were observed on flowers of W. densiflora at all of the sites, except Durban. Flies captured at the study sites carried thick encrustations of grey pollen on the upperside of their thorax. Subsamples of grains of this pollen, obtained by swabbing small cubes of fuschin gel over the thorax of the 13 flies captured at the Highover site, revealed that their pollen loads consisted almost entirely (median = 100%; range = %) of Watsonia pollen. Flies foraging on Watsonia inflorescences readily approached and probed Disa nervosa inflorescences that were proffered in the near vicinity. Of the 15 flies tested in this way, 12 (80%) flew to the orchid inflorescences and probed flowers. Even though the numbers of open flowers on Watsonia and Disa inflorescences were similar, flies probed c. 6.3 flowers on Watsonia inflorescences but just 1.9 flowers on the orchid inflorescences (Table 2). Another difference in behaviour that was clearly visible on video footage was that flies pumped 2 3 times into Watsonia flowers, whereas they inserted their proboscis just once into the Disa flowers. of the orchid are significantly shorter than those of the Watsonia (Table 2), and are produced singly, as opposed to the multiple (up to 10) inflorescences per plant of W. densiflora. In terms of spectral reflectance, flowers of the orchid and Watsonia are very closely matched (Fig. 3). The mean spectral reflectance of the orchid petals is most similar to that of the outer surface of the tepals of Watsonia and somewhat midway between the spectra of the outer and inner Watsonia tepals, which differ slightly in overall brightness and UV reflectance (Fig. 3). EFFECT OF WATSONIA ON DISA POLLINATION SUCCESS Mean levels of pollination deposition on orchid stigmas varied significantly among sites from zero at the heavily urbanized Durban site to almost 70% at the Weza site (F 3,69 = 20.1, P < 0.001; Table 3). Similar trends were evident for pollen removal from flowers (F 3,68 = 22.9, P < 0.001; Table 3). There was no significant relationship between mean pollen deposition (or removal) in individual plants of D. nervosa and proximity to Watsonia plants at either Highover (deposition: R 2 = 0.04, d.f. = 27, P = 0.30; removal: R 2 = 0.006, d.f. = 27, P = 0.69) or Josefsdal (deposition: R 2 = 0.05, d.f. = 18, P = 0.92; removal: R 2 = 0.001, d.f. = 18, P = 0.91). Other sites had too few Disa plants to conduct similar univariate regressions. There was also no significant effect of proximity to Watsonia on pollination success of the orchid (F 1,49 = 0.09, P = 0.75) when all sites were included as factors in an ANCOVA, with distance to Watsonia plants as a covariate. PLANT TRAITS The floral tube length in D. nervosa (c mm, Table 2) and W. densiflora (c mm, Table 2) closely matched the average proboscis length of P. aethiopica (c mm, see above) at the Highover site. Both orchid and Watsonia tended have a similar number of open flowers per inflorescence, but the inflorescences DISCUSSION The hypothesis that Disa nervosa is a mimic of flowers in the Watsonia densiflora complex was only partly supported by the results. As predicted by the mimicry hypothesis, the orchid shares the same habitat and flowering time as W. densiflora (and its almost morphologically identical close relative W. pulchra) and

6 276 S. D. JOHNSON and S. MORITA Figure 3. Mean spectral reflectance of flowers of Disa nervosa and Watsonia densiflora. Table 3. Mean (±SD) pollination and fruiting success of plants of Disa pulchra. Means within a column that differ significantly (see text for ANOVA results) are indicated by different letters. The Durban and Saddleback Pass populations were excluded from this analysis on account of their non-normal distribution and small sample size, respectively Site Number of plants examined Flowers pollinated (%) Pollinaria removed (%) Fruit set (%) Durban Highover ± 26.4 a 42.9 ± 21.2 a 41.0 ± 24.0* Josefsdal ± 19.2 b 20.1 ± 18.2 b Roselands ± 3.8 b 19.9 ± 11.9 b Saddleback Pass ± ± 11.1 Weza ± 21.9 a 70.4 ± 21.4 a Based on a sample of 45 plants has a remarkably similar distribution to these two Watsonia species (Fig. 2). In addition to the study sites, we have recorded D. nervosa growing in the same habitats as W. densiflora or W. pulchra at numerous places across the distribution range of the orchid. The orchid and W. densiflora are both pollinated by the long-proboscid fly Philoliche aethiopica, although we lack observations at sites where the orchid occurs with W. pulchra (Fig. 2). Flies readily visited plants of D. nervosa that either occurred naturally or were placed close to Watsonia inflorescences. This behaviour is not simply a result of generalist foraging as the observations and granular pollen load analysis indicate that the flies show high levels of fidelity to Watsonia flowers. Nocturnal observations were not carried out during this study, but additional pollination of the study species by moths seems unlikely given the absence of human-detectable floral scent in the evening. Flower colour and depth in D. nervosa and W. densiflora are closely matched (Fig. 3, Table 2). However, the floral morphology of D. nervosa does not match its putative model as closely as has been reported in other South African orchids that appear to be mimics (Johnson, 1994, 2000; Johnson et al., 2003a; Anderson et al., 2005). In particular, the orchid has relatively slender perianth segments and a shorter inflorescence than Watsonia (Fig. 1A D, Table 2). However, these differences in morphology may not be important for fly discrimination, as flies readily probed misshapen and damaged flowers of W. densiflora (Fig. 1F). Although the sympatric occur-

7 LYING TO PINOCCHIO 277 rence of D. nervosa with Watsonia plants at all of the sites, except Weza, is suggestive of an obligate dependence, there was no evidence that pollination success in the orchid is dependent on Watsonia. The lack of a significant effect of proximity to plants of W. densiflora on pollination success of the orchid could be attributed to lack of variation in fly conditioning over the small spatial scale (no orchids were found more than 300 m from plants of W. densiflora at any of the sites where both species were present). However, the high levels of pollination success of the orchid at the Weza site, where no Watsonia plants were present in the near vicinity, suggests that nonconditioned flies are also likely to visit its flowers. Unfortunately, although we caught one P. aethiopica here, we were unable to obtain any data on the abundance of tabanid flies at this site or establish whether Watsonia populations were present within pollinator flight-distance of this orchid population. Our idea that the pink flowers of D. nervosa are adapted to resemble those of Watsonia densiflora would need to be tested within a phylogenetic framework in order to establish an association of traits and their function (Harvey & Pagel, 1991; Johnson et al., 2003a). A phylogeny for Disa section Emarginatae is currently lacking, but the large size and pink colour of flowers in D. nervosa are unique features within this group. Assuming that the group is monophyletic and that the pink flowers and large size of D. nervosa are not basal within it, this suggests that these features may be adaptations for the current pollination system of D. nervosa and not simply plesiomorphic traits carried over from its ancestors by descent. Interestingly, resemblance of pink Watsonia flowers pollinated by the fly P. aethiopica appears to have evolved independently in two distantly related sections of Disa. Additionally, these two instances may represent different points along a spectrum from generalized food deception to Batesian mimicry. On the one hand, as described in this study, D. nervosa (section Emarginatae) has flowers which are very similar in colour and shape to those in the W. densiflora complex, but it differs from that complex in its inflorescence structure (the orchid has shorter inflorescences produced singly, as opposed to the taller multiple inflorescences of W. densiflora). Disa pulchra Sond. (section Stenocarpa), on the other, closely resembles Watsonia lepida N.E.Br. in both floral traits and the possession of a single inflorescence (Johnson, 2000). The peak flowering of the Watsonia species in the two systems is about two months apart (W. lepida in December and the W. densiflora complex in February) and they have largely nonoverlapping geographical distributions. These differences in flowering time and distribution of the two Watsonia species are closely matched by the two orchids In conclusion, while most of the evidence in this paper is consistent with the idea that D. nervosa is pollinated by the tabanid fly P. aethiopica by virtue of mimicry of the flowers of the rewarding irid W. densiflora, a role for generalized food deception in the orchid cannot be excluded on account of its high level of pollination success at one site where W. densiflora plants were absent (at least from the near vicinity). Several orchid species pollinated by long-proboscid flies have been proposed as Batesian mimics of sympatric rewarding plants, yet we know almost nothing about the degree to which foraging choices by these flies are influenced by innate vs. conditioned preferences. Because a key assumption underlying any floral mimicry hypothesis is that the mimic gains benefit primarily from conditioned behaviour of the pollinator, conditioning experiments similar to those that have been conducted on bees (cf. Gigord et al., 2002) urgently need to be carried out on these flies in order to better understand the evolution of floral traits in the plants they pollinate. ACKNOWLEDGEMENTS We thank Brian Spitzer for help in the field, Bruce Anderson for advice about field sites, the managers of Roselands and Highover farms and Weza-Ngele Forest and Umtamvuna Reserve for access, and KZN Wildlife for permits. Work by SIM was supported by the National Science Foundation under Dissertation Enhancement Grant No. INT , and a J. William Fulbright Student Scholarship from the U.S. Department of the State REFERENCES Anderson B, Johnson SD The effects of floral mimics and models on each others fitness. Proceedings of the Royal Society B Biological Sciences 273: Anderson B, Johnson SD, Carbutt C Exploitation of a specialized mutualism by a deceptive orchid. American Journal of Botany 92: Beattie AJ A technique for the study of insect-borne pollen. Pan-Pacific Entomologist 47: 82. Dafni A Pollination in Orchis and related genera: evolution from reward to deception. In: Arditti J, ed. Orchid biology: a review and perspectives. Ithaca, NY: Comstock Publishing Associates/ Cornell University Press, Dafni A, Ivri Y Floral mimicry between Orchis israelitica Baumann and Dafni (Orchidaceae) and Bellevalia flexuosa (Liliaceae). Oecologia 49: Gigord LDB, Macnair MR, Stritesky M, Smithsonian A The potential for floral mimicry in rewardless orchids: an experimental study. Proceedings of the Royal Society B 269: Goldblatt P The genus Watsonia. Annals of the Kirstenbosch Botanical Gardens 19:

8 278 S. D. JOHNSON and S. MORITA Goldblatt P, Manning JC The long-proboscid fly pollination system in southern Africa. Annals of the Missouri Botanical Garden 87: Gumbert A Color choices by bumble bees (Bombus terrestris): innate preferences and generalization after learning. Behavioral Ecology and Sociobiology 48: Gumbert A, Kunze J Colour similarity to rewarding model plants affects pollination in a food deceptive orchid, Orchis boryi. Biological Journal of the Linnean Society 72: Harvey PH, Pagel MD The comparative method in evolutionary biology. Oxford: Oxford University Press. Jersakova J, Johnson SD, Kindlmann P Mechanisms and evolution of deceptive pollination in orchids. Biology Reviews 81: Johnson SD Evidence for Batesian mimicry in a butterfly-pollinated orchid. Biological Journal of the Linnean Society 53: Johnson SD Batesian mimicry in the non-rewarding orchid Disa pulchra, and its consequences for pollinator behaviour. Biological Journal of the Linnean Society 71: Johnson SD, Alexandersson R, Linder HP. 2003a. Experimental and phylogenetic evidence for floral mimicry in a guild of fly-pollinated plants. Biological Journal of the Linnean Society 80: Johnson SD, Peter CI, Nilsson LA, Ågren J. 2003b. Pollination success in a deceptive orchid is enhanced by co-occurring nectar plants: evidence for the magnet species effect. Ecology 84: Kunze J, Gumbert A The combined effect of color and odor on flower choice behavior of bumble bees in flower mimicry systems. Behavioral Ecology 12: Linder HP Taxonomic notes on the Disinae III. A revision of Disa Berg. excluding sect. Micranthae. Contributions from the Bolus Herbarium 9: Nilsson LA Mimesis of bellflower (Campanula) by the red helleborine orchid Cephalanthera rubra. Nature 305: Thomson JD Effects of variation in inflorescence size and floral rewards on the visitation rates of traplining pollinators of Aralia hispida. Evolutionary Ecology 2: Verdoorn IC Watsonia densiflora. Flowering Plants of Africa 33: tabs 1293 & 1294.