Floral specialization for beetle pollination and its implications for pollen dispersal in an African orchid

Abstract

Premise

Pollination by beetles is relatively rare in orchids, and this has been attributed to the clumsy behavior of beetles being unsuitable for the precise pollen transfer mechanisms that characterize the orchid family. We investigated floral specialization for beetle pollination in the rare fire‐dependent South African orchid Disa elegans and explored its implications for the efficiency and spatial pattern of pollen dispersal.

Methods

We observed flower visitors and identified their pollen loads. We studied floral traits, including spectral reflectance patterns, nectar secretion, and scent chemistry. We tracked the dispersal of color‐labeled pollen.

Results

Disa elegans was found to be pollinated by large scarab beetles. Apparent floral adaptations for beetle pollination include the platform‐like corymbose inflorescence of upward‐facing, bowl‐shaped flowers, secretion of very dilute nectar on exposed surfaces of the petals, and fruity floral scent dominated by the monoterpene alcohol R‐(–)‐β‐linalool and benzenoid ester methyl benzoate. Beetles carry large loads of pollinaria and transfer ~13% of the pollen they remove from anthers to stigmas. We found a classic leptokurtic kernel of pollen dispersal with an average distance from donors to recipients of 6.7 m. Self‐pollen made up ~30% of all pollen deposited on stigmas by beetles. These pollen dispersal patterns are similar to those obtained in plants pollinated by other insect groups, such as bees.

Conclusions

These results provide evidence of floral specialization for beetle pollination in an orchid species and show that beetles can be effective agents of pollen dispersal in orchid populations.

Beetles were one of the first insect groups to form pollination mutualisms with plants (Bao et al., 2019), and beetles still play a major role in pollination (Bernhardt, 2000). They are generally considered clumsy fliers that require bowl‐shaped flowers in which to land, and their lack of long mouthparts means that they can feed only on readily accessible pollen and nectar. Many also feed on floral tissues (Bernhardt, 2000; Gottsberger et al., 2012). These traits make them seem unlikely as pollinators of orchids, which have precise “lock and key” pollination mechanisms. Van der Pijl and Dodson (1966), for example, reached the conclusion that there was “no trend toward adaptation to beetles as pollinators” in the Orchidaceae.

Despite beetles being the most species‐rich order of insects (Stork et al., 2015) and orchids being, by some accounts, the most species‐rich plant family, these two mega‐lineages appear to have had relatively limited interactions over their evolutionary history. Structures resembling orchid pollinaria (pollen packages) have been found attached to weevils and toe‐winged beetles preserved in amber that is >20 million years old (Poinar, 2016), but this by itself does not demonstrate that these orchids had specific pollination associations with beetles. One aspect of beetles that does make them suitable as pollinators of orchids is their smooth carapace, which forms a good surface for adhesion of the glue‐like viscidia that are part of the pollinaria of most orchids. Studies conducted since the publication of Van Der Pijl and Dodson (1966) have provided evidence that beetles do pollinate orchids and that some are evolutionarily specialized for this mode of pollination (Bernhardt, 2000). Despite their reputation as “mess and soil” pollinators (Faegri and Van Der Pijl, 1979), chafer beetles (Cetoniinae) have been shown to be the primary pollinators of the orchids Satyrium microrrhynchum Schltr., Orthochilus welwitschii Rchb.f., and O. ensatus Lindl. (Bytebier) (Peter and Johnson, 2009); Eulophia parviflora (Lindl.) A.V.Hall (Peter and Johnson, 2014); and O. ruwenzoriensis (Rendle) Bytebier (Singer and Cocucci, 1997). In the Asian genus Luisia, chafer beetle pollination has been recorded for L. curtisii Seidenf. (Pedersen et al., 2013) and L. teres Lindl. (Arakaki et al., 2016; Sugiura et al., 2020), with some reports indicating sexual deception of male beetles in some populations of the latter species (Arakaki et al., 2016). In South Africa, another group of scarabs, the Hopliinae, is responsible for pollination of the orchid Ceratandra grandiflora Lindl. (Steiner, 1998). While these systems are relatively specialized in the sense that pollen transfer among flowers is effected exclusively or nearly exclusively (>80%) by beetles, there are also cases where beetles form part of pollinator assemblages for orchids with more generalized pollination systems such as Neottia (formerly Listera) ovata (L.) Hartm. in Europe (Nilsson, 1981) and Disa fragrans Schltr. in Africa (Johnson and Hobbhahn, 2010).

Floral trait specialization for beetle pollination in orchids has not been well studied. In general, orchids adapted for beetle pollination have inflorescences with flowers arranged as a landing platform or have flowers with a large platform‐like labellum and very short floral spurs or no spurs at all. Nectar is generally reported to be very dilute (~10% sugar concentration) in flowers pollinated by cetoniine beetles (Johnson et al., 2007; Steenhuisen and Johnson, 2012; Arakaki et al., 2016). Floral scent chemistry in beetle‐pollinated plants is diverse, and studies have identified responses of antennae of cetoniine beetles to monoterpene alcohols such as linalool, as well as to benzenoid esters such as methyl benzoate (Donaldson et al., 1990; Johnson et al., 2007).

One of the key questions about beetle pollination in orchids is whether it is efficient with respect to pollen transfer among flowers. Clumsy crawling behavior, imprecise positioning of their bodies with respect to floral structures, and long residency times of beetles in flowers would be expected to lower the overall fraction of pollen that reaches stigmas and increase the risk of pollinator‐mediated self‐pollination. Some beetle‐pollinated orchids have mechanisms such as slow pollinarium reconfiguration and anther cap retention that are considered to mitigate risks of geitonogamy (Peter and Johnson, 2006a, 2006b). Data on pollen dispersal by beetles is very limited and is mostly based on indirect methods, such as mark‐recapture experiments (Young, 1988; Garcia‐Robledo, 2010) or paternity analysis (De Almeida et al., 2018), and not on direct tracking of pollen flow. The general trend of pollen dispersal in orchids is for dispersal kernels to be strongly leptokurtic and right‐skewed, with a substantial fraction (~35%) of the pollen deposited on the donor plant's own stigmas through pollinator‐mediated geitonogamy and within‐flower self‐pollination (Hobbhahn et al., 2017).

Previous observations of the Cape orchid Disa elegans Sond. Ex. Reichb.f. made by various naturalists led us to hypothesize that this species is adapted for pollination by beetles. Bolus (1893) illustrated a hopliine scarab beetle with pollinaria of D. elegans; Steiner (1998) and Liltved and Johnson (2012) cited their unpublished observations of the chafer beetle Trichostetha signata (Fabricius) visiting flowers of this orchid. We conducted a detailed study to address the hypothesis of beetle pollination in this species. Specifically, we addressed the following questions: (1) Are beetles the primary pollinators of D. elegans? (2) Are floral traits of the orchid, such as shape, nectar rewards, and scent chemistry, consistent with beetle pollination? (3) Do beetles transfer pollen effectively among flowers, and what is the shape of the pollen dispersal kernel? (4) What are the rates of pollinator‐mediated self‐pollination, and does self‐fertilization lead to a reduced proportion of seeds with viable embryos?

MATERIALS AND METHODS

Study species and study sites

Disa elegans occurs in marshes, in swamps, and along stream banks in the inland mountains between Clanwilliam and Oudtshoorn in the Western Cape, South Africa. The species is known from relatively few populations and flowers only after fire (Liltved and Johnson, 2012). The 10–20 yr intervals between fires in Cape fynbos vegetation mean that opportunities to study the pollination of this species are very infrequent. Plants produce a platform‐like corymbose inflorescence with upward‐facing white flowers with maroon and yellow bands on the tips of the lateral petals and lip (Figure 1A, B). Double resupination of the ovary orients the lip toward the center of the inflorescence (Figure 1B). In November–December 2009, we studied a population of ~2300 flowering plants in a recently burned wetland area of ~15 ha in the Groot Winterhoek Mountains (32 59′20 S, 19 04′01 E; elevation 938 m). A voucher specimen from the study population (Hobbhahn 012, NU 00373‐9) was deposited in the Bews herbarium in Pietermaritzburg.

Figure 1.

Disa elegans and its flower visitors. (A) Flowering plants in wetland habitat. (B) Inflorescence. (C) Female chafer beetle Trichostetha capensis carrying a large load of pollinaria. (D) Chafer beetle T. signata with large load of pollinaria feeding from nectar on the petals. (E) Hopliine beetle Lepithrix sp. feeding on nectar. (F) Ant Camponotus niveosetosus feeding on nectar. (G) Side view of a dissected flower with the sepals removed, showing nectar droplets on the lip and petals (p = petal, a = anther, r = rostellum, s = stigma, l = labellum). Scale bars: A = 10 cm, B = 10 mm, C = 5 mm, D = 5 mm, E = 10 mm, F = 5 mm, G = 5 mm.

Flower visitors

Observations of flower visitors were conducted for ~42 h over a period of 7 d in November 2009. We noted insect visitors, their behavior on the inflorescences, and whether or not they carried pollinaria. A sample of insect visitors was collected for identification, body length measurements, and quantification of pollen loads. Insect specimens are deposited at the University of KwaZulu‐Natal, Pietermaritzburg, and the University of Stellenbosch.

Floral morphology and spectral reflectance

We measured plant height and counted the number of flowers produced per inflorescence. Length and width of each flower, dorsal sepal, lateral sepal, and lip were measured with digital calipers. Spectral reflectance of various flower parts (dorsal sepal, lateral petals, and labellum) as well as the maroon patches were measured using an Ocean Optics USB 2000 spectrometer (Ocean Optics, Duiven, Netherlands) with the configuration described by Johnson and Andersson (2002).

Floral nectar

The volume of the standing crop of nectar in 27 flowers was measured at midday with calibrated 5 µL microcapillary tubes (Blaubrand, Wertheim, Germany), and sugar concentration (in sucrose equivalents) was measured with a 0%–50% hand refractometer designed for small nectar volumes (Bellingham & Stanley model 45‐81, Tunbridge Wells, UK). We also measured nectar volume and concentration in 19 flowers following 24 h visitor exclusion with fabric bags. Nectar removal with microcapillary tubes from flat, exposed surfaces on petals and lip may be incomplete because thin liquid layers may not be taken up by the tubes. We therefore also used filter paper wicks as collection devices to assess the amount of sugars in nectar secreted over 24 h in an additional 15 flowers bagged on a hot, dry, windless day and 30 flowers bagged on a cool, windy day. While nectar collection with filter paper does not enable assessment of nectar volume, it may provide a more accurate representation of the amount of sugar secreted over 24 h. Filter paper samples were air dried, eluted in distilled water, and filtered with a 0.45 micron syringe filter. Filtered samples were analyzed using a Shimadzu HPLC (LC‐20AT; Shimadzu, Kyoto, Japan) equipped with a differential refractometric detector (RID10A) and a Phenomenex column (Rezex RCM‐Monosaccharide, 200 × 780 mm 8 micron; Phenomenex, Torrance, California, USA). The elution was isocratic, using ultrapure water as the mobile phase. Known amounts of sucrose, glucose, and fructose were run with the samples to enable identification and quantification of these common nectar sugars.

Floral scent analysis

We sampled scent of flowering plants at the Grootwinterhoek population by enclosing 10 inflorescences individually in polyacetate bags and pumping air from these bags through cartridges containing adsorbent polymer for 30 min at a realized flow rate of 50 mL/min. Cartridges were filled with a mixture of 1 mg of Tenax and 1 mg of Carbotrap activated charcoal. We analyzed seven samples using a Varian CP‐3800 GC (Varian, Palo Alto, California, USA) with a 30 m × 0.25 mm internal diameter (film thickness 0.25 µm) Alltech EC‐WAX column (Alltech, Nicholasville, Kentucky, USA) coupled to a Varian 1200 quadrupole mass spectrometer in electron‐impact ionization mode. An additional three samples were analyzed using the same instrument with a DB5 column to improve compound identification by obtaining Kováts retention indices for a column with a different phase. Cartridges were placed in a Varian 1079 injector equipped with a “Chromatoprobe” thermal desorption device. Helium at a flow rate of 1 mL min⁻¹ was used as the carrier gas. The injector was held at 40°C for 2 min with a 20:1 split, increased to 200°C at 200°C min⁻¹ in splitless mode for thermal desorption, and held there for 10 min. After a 3 min hold at 40°C, the GC oven was ramped up to 240°C at 10°C min⁻¹ and held there for 12 min. Compounds were identified using Varian Workstation software with the NIST05 mass spectral library (NIST, Gaithersburg, Maryland, USA) and verified, where possible, using retention times of synthetic standards and comparisons of obtained Kováts indices to those published for polar and non‐polar columns. Compounds present at similar abundance in the controls were considered contaminants and excluded from further analysis.

Optical isomers (enantiomers) of linalool were separated on a SCION‐Chirasil‐DEX enantioselective column (25 m × 0.25 mm ID, 0.25 μm film thickness) (Scion instruments, Goes, Netherlands). The retention time for solvent‐eluted floral samples of linalool was compared to retention times for injections of rac‐(±)‐linalool (Sigma‐Aldrich) and R‐(–)‐linalool (Fluka, Sigma‐Aldrich, Merck, Darmstadt, Germany). Floral samples were diluted until the size of the linalool peak approximately matched those of the enantiomer standards.

Pollination success and pollen transfer efficiency

Pollination success in the study population was assessed at peak flowering with two methods. A total of 68 inflorescences were examined and we recorded the number of open flowers and the number of pollinated flowers for each inflorescence. In another 20 plants, one arbitrarily chosen flower per inflorescence was checked for pollinarium removal and pollen deposition. We recorded both the mean number of pollinia removed (m _r) and the mean number of massulae deposited on the stigma (m _s) using a 20× hand lens. We also counted the number of massulae per pollinium (m) from 20 flowers by dabbing pollinia onto sticky tape until all massulae were spread out and separated. Pollen‐transfer efficiency (PTE) was estimated from the equation PTE = m _s/(m × m _r) (Johnson et al., 2005). We also calculated the fraction of massulae produced per flower that reaches stigmas as m _s/(2 × m) (Harder, 2000).

Pollen dispersal kernels

We tracked the dispersal of pollen from individuals in the population by color labeling pollen with histochemical stains (Peakall, 1989). We stained 350 pollinaria in the anthers of 182 flowers on 23 plants using a 10 µL syringe (Hamilton, Reno, Nevado, USA). We used the following histochemicals: rhodamine pink (0.2%), fast green (1%), and gentian violet (premixed medicinal preparation “Alpha”). Previous studies of orchids have shown that these stains do not influence pollen removal or deposition, including patterns of pollen carryover (Johnson et al., 2004, 2005). We added a surfactant (TWEEN 20), which facilitates stain penetration of hydrophobic pollinia. We selected three plants for each “focal” group, each stained with different colors, and replicated these “focal” groups such that each group was separated by ~50 m to minimize overlap in dispersal radii of pollen from donors stained with histochemical stain of the same color. After staining, plants were left exposed to pollinators for 3 d. We then checked all stained flowers for removal, and the stigmas of all open and wilting flowers, including donor plants, in the population for deposition of stained pollen. Using a 20× hand lens, we counted the number of stained and unstained massulae deposited on each flower that showed evidence of deposition of stained pollen. We also measured the distance to the nearest donor of the same stain color, and counted the intervening plants between the pollen donor and recipient that did not receive pollen of that color in order to calculate how many plants were “skipped” during pollen dispersal. Hobbhahn et al. (2017) used summary values from this pollen‐labeling experiment in a meta‐analysis of pollen fates in rewarding versus rewardless orchid species, but the specific results for D. elegans are reported here for the first time.

Breeding system

We hand pollinated flowers with self‐ and cross‐pollen to assess the degree of self‐compatibility and predispersal inbreeding depression. Two virgin flowers in similar inflorescence positions were chosen in each of 25 bagged inflorescences and assigned to be either self‐ or cross‐pollinated. Self‐pollen was taken from the same flower. Cross‐pollen was taken from inflorescences 10–15 m from the recipient plant, with one donor per recipient. Pollinia were dabbed onto the stigma until the stigma was saturated with massulae. Following manual pollination, inflorescences were bagged until seedpod maturity. Unpollinated flowers enclosed in the mesh bags served as a test for autonomous selfing capacity. We recorded the time taken for flowers to wilt after pollination, as this could be a consequential outcome of self‐ versus cross‐pollination (Gilissen, 1977). The resulting seedpods were collected in January 2010, and ~400 seeds/pod were examined microscopically under backlighting to establish the proportion that had developed embryos.

Statistical analysis

Unless otherwise stated, data were analyzed using generalized linear models (GLMS) implemented in SPSS 25 (IBM, Armonk, New York, USA). We used Gaussian, binomial, and negative binomial models with their canonical link functions to analyze measurement, proportion, and count data, respectively. Models were adjusted for overdispersion when required, and significance testing was based on likelihood ratios. For analysis of the proportion of ovules that developed into viable seeds for flowers that were either self‐ or cross‐pollinated on each plant, we used a paired t‐test based on logit‐transformation of the proportion values. Data that did not conform to known distributions were analyzed using nonparametric statistics. Unless indicated otherwise, data are summarized as means (±SE).

RESULTS

Flower visitors

The flowers of D. elegans were pollinated almost exclusively by scarab beetles: two fruit chafer species—Trichostetha capensis (L) (Figure 1C) and Trichostetha signata (Figure 1D)—and an unidentified hopliine species, Lepithrix sp. (Figure 1E). Overall body length (mm) of the fruit chafers was twofold greater than that of the hopliine beetle (16.6 ± 1.60 vs. 7.9 ± 0.16, χ ² = 187.41, P < 0.001), but there was no significant difference in body length between the two fruit chafer species (16.4 ± 0.36 vs. 16.7 ± 0.25, χ ² = 0.43, P = 0.51). The flowers are pollinated either when beetles lap nectar from the surface of the petals or when they face in the opposite direction to lap nectar from the lip. In both of these feeding positions, the ventral surface of their thorax comes into contact with the rostellum arms, which bear the viscidia (Figure 1G). Pollinaria are thus attached to the ventral surface of the thorax and dangle directly onto the stigma. The beetles did not feed on the perianth tissue.

We examined a total of 718 flowering plants of D. elegans and recorded 41 individual fruit chafers on flowers, 37 of which carried pollinaria; and 46 hopliine scarabs, 26 of which carried pollinaria. The proportion of individuals carrying pollinaria was higher for fruit chafers than for hopliine scarabs (0.90 ± 0.046 vs. 0.65 ± 0.070, χ ² = 8.15, P = 0.04). Fruit chafers also carried nearly fourfold more pollinaria, on average, than did hopliine scarabs (10.1 ± 1.65 vs. 2.30 ± 0.41, χ ² = 36.9, P < 0.001), but the number of pollinaria carried did not differ significantly between the two fruit chafer species (χ ² = 0.35, P = 0.55). Some fruit chafer individuals carried ≤40 pollinaria (Figure 1C). Chafer beetles flew distances of ≤15 m between plants when foraging, and the most active individuals spent a median of just 6 s feeding per flower (n = 7; range: 5–10 s). However, some beetles rested in flowers and spent ≤12 min on a single inflorescence before flying to another plant. Male:female ratios of individuals of the fruit chafers captured on flowers of D. elegans were 14:5 for T. signata and 6:3 for T. capensis. The Lepithrix beetles lack obvious sexual dimorphism and were not sexed.

Aside from the beetles, we recorded nectar feeding by nine calliphorid flies, 64 ants (mostly Camponotus niveosetosus Mayr; Figure 1F), one pompilid wasp, and three window‐winged moths. None of these visitors carried pollinaria, aside from the pompilid wasp, which carried four pollinaria, and one of the window‐winged moths, which carried a single pollinarium.

Floral morphology and spectral reflectance

The inflorescences of a sample of 20 plants in this population were 52.10 ± 2.35 cm tall, with 12.3 ± 1.23 flowers per inflorescence, of which 7.95 ± 0.49 were open simultaneously. The flowers are shallowly bowl‐shaped and measured 32.03 ± 1.31 × 18.35 ± 0.61 mm (length × width). Details of all floral morphological traits measured are given in Appendix S1.

The flowers are largely white and absorb UV light, but the vivid yellow tips of the petals and the creamy‐yellow tips of the lip (median petal) are strongly UV reflecting (Figure 2). The UV‐reflecting yellow tips of the petals are flanked by maroon patches that are UV absorbing (Figure 2).

Figure 2.

Spectral reflectance of different parts of the perianth of Disa elegans.

Floral nectar

Nectar is secreted from the surface of the maroon patches on the petals and lip, forming a film of droplets (Figure 1G) that is accessible to short‐tongued insects. The mean (± SE) standing crop of nectar per flower (n = 27) at midday was 0.30 ± 0.13 µL with a sugar concentration of 4.20 ± 0.71 g per 100 g, corresponding to 0.01 ± 0.005 mg sugar per flower. The standing crop of nectar on the petals of the 27 flowers sampled was 0.27 ± 0.12 µL, and on the lip it was 0.04 ± 0.01 µL (χ ² = 3.33, P = 0.068). The amount of nectar produced per flower (n = 19) after 24 h was 1.53 ± 0.21 µL, with a sugar concentration of 5.06 ± 0.37 g per 100 g, corresponding to 0.07 ± 0.01 mg sugar per flower. Using HPLC analysis of nectar collected on filter paper, we derived two additional estimates of 24 h sugar production per flower: 0.04 ± 0.01 mg (n = 15) on a hot day and 0.16 ± 0.02 mg (n = 30) on a cooler day. The mean (±SE) percentages of the sugars in D. elegans nectar (n = 30 samples) were as follows: sucrose = 74.2 ± 0.04%, fructose = 11.2 ± 0.02%, and glucose = 14.5 ± 0.02%.

Floral scent chemistry

The flowers of D. elegans have a strong fruity scent to the human nose. The scent consists of a relatively simple blend of compounds dominated by the monoterpene alcohol linalool and a benzenoid ester, methyl benzoate (Table 1). The enantiomer of the former compound was identified as R‐(–)‐linalool (Appendix S2). The overall emission rate of volatiles per inflorescence was 2492.1 ng h⁻¹, and the rate per flower was 277.2 ng h⁻¹.

Table 1.

Chemical composition of the floral scent of Disa elegans. Values are percentages of total peak area (excluding contaminants) followed by the number of samples in which the compound was identified (results >10% are in bold). Linear Kováts retention indices (for samples run on a polar column) were calculated relative to retention times of n‐alkanes. Identification (ID) criteria: A = mass spectral library match and Kováts retention index consistent with published values; B = Kováts retention index for samples run on a non‐polar column consistent with published values; C = retention time and mass spectrum compared to that of synthetic standards injected under identical conditions to samples.

Compound	Linear Kováts retention index	ID criteria	Mean % ± SE (occurrence), N = 7
Aliphatics
Nonanal	1410	A	1.79 ± 0.92 (5)
Decanal	1516	A	0.64 ± 0.42 (3)
Benzenoids
Benzonitrile	1635	A	2.09 ± 1.92 (4)
Methyl benzoate	1650	ABC	29.4 ± 6.09 (7)
3,5‐Dimethoxytoluene	1878	ABC	0.01 ± 0.01 (1)
Benzyl alcohol	1899	ABC	0.07 ± 0.03 (4)
p‐Cresol	2100	AC	0.06 ± 0.04 (2)
Irregular terpenes
Geranyl acetone	1872	A	0.17 ± 0.11 (5)
Monoterpenes
β‐Myrcene	1201	BC	0.33 ± 0.25 (3)
Limonene	1228	BC	0.78 ± 0.72 (2)
(E)‐Linalool oxide (furanoid)	1455	A	0.12 ± 0.04 (5)
(Z)‐Linalool oxide (furanoid)	1484	A	0.1 ± 0.05 (3)
R‐(–)‐β‐Linalool	1556	ABC	57.62 ± 10.53 (7)
Hotrienol	1624	A	0.08 ± 0.04 (4)
Geranyl acetate	1772	AB	4.72 ± 4.1 (7)
Geraniol	1866	AB	1.17 ± 0.93 (7)
3,7‐Dimethyl‐1,5‐octadien‐3,7‐diol	1956	A	0.14 ± 0.1 (5)
2,6‐Dimethyl‐1,7‐octadiene‐3,6‐diol	2128	A	0.07 ± 0.07 (2)
Unknown sesquiterpenes
204*,161,133,105,189,91,146,175,107,79,119	1760		0.61 ± 0.29 (6)
204*,161,122,107,105,79,93,91,162,55,76	1796		0.04 ± 0.03 (3)

Pollination success and pollen transfer efficiency

Pollination success of D. elegans was remarkably high. In the first survey of 375 flowers on 68 plants, we found that 54 ± 5.0% of flowers per plant were pollinated. Both the number of pollinated flowers and the overall proportion of flowers pollinated on inflorescences were positively correlated with overall display size (Appendix S2). In the second survey of a single flower on each of 20 plants, we found that 85% of flowers had at least one pollinarium removed and 90% of flowers were pollinated. Each pollinium had 1076 ± 177.1 massulae. The mean number of pollen massulae on stigmas was 231.9 ± 26.15, and the mean number of pollinaria removed per flower was 1.6 ± 0.16. This translates to an overall pollen transfer efficiency of 10.8% of produced pollen being exported to stigmas, and 13.5% of removed pollen being exported to stigmas.

Pollen dispersal

We stained 350 pollinaria with an estimated 376,600 massulae and located just 7548 (2%) of these stained massulae on stigmas. This fraction is much lower than the value of 10.8% that we calculated for overall pollen transfer efficiency in the population, indicating that some of the stained pollen may have been exported outside the search perimeter. This could explain why 72% of the stained pollen that we located on stigmas was found deposited on the source plants whereas just 28% was found deposited on other plants. Self‐pollen made up 30.4 ± 6.9% of all pollen massulae deposited on the stigmas of plants that we stained, and this provides a measure of the overall rate of self‐pollination in the population. Beetles skipped an average of 3.33 intervening plants between donors and recipients (range: 0–12; Figure 3A). Dispersal kernels showed a much higher probability of dispersal to the source‐plant in the case of pollen massulae than in the case of recipient plants (Figure 3B–C). The average distance from source to recipient plants was 7.82 m (median = 4.7 m, range: 0.05–26 m; Figure 3B).

Figure 3.

Dispersal of stained pollen of Disa elegans. (A) The number of plants skipped (i.e., not receiving stained pollen) between donor and recipient plants. (B) Distances between donor and recipient plants. (C) Distances of dispersal of individual pollen massulae. Dark‐shaded bars represent donor plants (i.e., self‐pollination); light‐shaded bars represent recipient plants.

Breeding system

Flowers of D. elegans are not capable of mechanical self‐pollination (pollinia in unvisited flowers remain in the anther) and therefore depend on pollinators. All of the flowers that were not pollinated remained open for ≥15 d (we did not record flower longevity beyond 15 d), but flowers wilted within a median of 3.14 d when hand pollinated (Mann‐Whitney U = 380.0, P < 0.0001). We found no difference in the time (in days) taken for self‐ and cross‐pollinated flowers to wilt (3.01 ± 0.21 vs. 3.06 ± 0.19, χ ² = 0.39, P = 0.84).

Fruits developed from 100% of self‐ and cross‐pollinated flowers on 25 plants, whereas no fruits developed from flowers that were bagged without manipulation of those plants (P < 0.0001, Fisher exact test). The mean percentage (lower SE, upper SE) of ovules that developed into seeds with embryos, after back‐transformation from the logit scale, was 66.6 (6.41, 7.53) for self‐pollinated and 70.1 (6.08, 7.18) for cross‐pollinated flowers (t = 0.442, P = 0.66), indicating that the species is fully self‐compatible.

DISCUSSION

Our results demonstrate the existence of a highly specialized beetle pollination system in D. elegans. Specialization is both ecological (a high degree of dependence on just three beetle species for pollination) and evolutionary (uniquely modified floral traits that play a functional role in the beetle pollination system). The primary pollinators were two Trichostetha chafer beetle species. The two Trichostetha species recorded here are known to visit the flowers of numerous species in the Cape Region, including many Protea species (Myburg and Rust, 1975; Johnson et al., 2012), but this is the first study to find that they are primary pollinators of a plant species. Our results are consistent with previous anecdotal observations of visits by cetoniine and hopliine scarabs to D. elegans flowers at other sites (Bolus, 1893; Steiner, 1998; Liltved and Johnson, 2012). The only other confirmed cases of specialized beetle pollination in the large orchid genus Disa are a system of sexual deception involving longhorn beetles in D. forficaria Bolus (Cohen et al., 2021) and a system of food‐deception involving pollen‐seeking ruteline beetles in D. similis Summerh. (Adit and Johnson, 2025). However, the pollination system of D. elegans is very different to these other two examples as it involves provision of nectar rewards and thus a very different set of floral traits.

A critical feature of chafer beetle pollination systems is that if nectar is offered as a reward, it must be accessible to the very short mouthparts of these insects (Ollerton et al., 2003; Johnson et al., 2007). Chafer beetles feed with sweeping motions of their maxillary palps. In D. elegans the droplets are formed on the flat surfaces of the lip and petals, making them readily accessible to chafer beetles. In other Disa species, nectar is secreted from the (usually spurred) dorsal sepal; hence, the secretion of nectar on the lip and petals in D. elegans is an evolutionarily novel strategy in the genus (Hobbhahn et al., 2013). In the chafer‐pollinated orchid Satyrium microrrhynchum nectar is presented as accessible droplets on hairs on the lateral petals at the base of the mouth of a shallow labellum cavity (Johnson et al., 2007). Interestingly, there is a clear global trend for the nectar of chafer‐pollinated flowers to be extremely dilute. We recorded a sugar concentration of 4%–5% for nectar of D. elegans, which is consistent with previous reports of extremely low nectar sugar concentration in chafer‐pollinated plants: 7%–8% in S. microrrynchum (Johnson et al., 2007), 2%–5% in Luisia teres (Arakaki et al., 2016), and 4%–10% in Protea species (Steenhuisen and Johnson, 2012). However, chafer beetles are not constrained to drinking dilute nectar; higher concentrations (12%–72%) have been recorded for nectar in some milkweed flowers pollinated by chafer beetles (Ollerton et al., 2003; Shuttleworth and Johnson, 2008).

The extremely dilute nectar of D. elegans may function as a filter that discourages visits by insects that are morphologically unsuited as pollinators. Although we recorded some visits by ants, flies, and pompilid wasps, other insects such as bees were notably absent. Studies of honeybees have shown that they strongly prefer nectar with a concentration of 30%–50% (Waller, 1972) and tend to avoid very dilute nectar altogether (Nicolson and Nepi, 2005).

Both color and scent have been shown to influence alighting behavior of chafer beetles (Schmera et al., 2004; Steenhuisen et al., 2013). Chafer beetles have been recorded to consume pollen of other plant species (Johnson and Nicolson, 2001), and the yellow markings on flowers of D. elegans could be interpreted as imitating anthers and pollen, as was recently suggested for UV‐absorbing yellow markings on the tips of the petals of the non‐rewarding beetle‐pollinated congener D. similis (Adit and Johnson, 2025). Lunau et al., (2017) has argued that contrasting color markings on flowers often function as a form of pollen mimicry that attracts flower‐visiting insects that visually assess the presence of pollen. However, pollen is typically UV absorbing (Lunau, 1995), whereas the yellow markings of D. elegans flowers are UV reflecting (Figure 2), thus weakening the argument that these yellow markings function as pollen imitation. The yellow and maroon markings of D. elegans coincide with the site of nectar production, making it seem more plausible that these markings are classical floral “nectar guides” (Sprengel, 1793).

Scent is considered an important trait for attraction of chafer beetles, and some flowers pollinated by these beetles are highly cryptic, suggesting that scent can be the primary cue for attraction (Johnson et al., 2007). Previous studies have shown antennal electrophysiological and behavioral responses of the South African chafer beetle Atrichelaphinis tigrina (Olivier) to several of the compounds emitted by D. elegans, including linalool, linalool oxide (furanoid), and methyl benzoate (Johnson et al., 2007; Steenhuisen et al., 2013). The European chafer Pachnoda marginata (Drury) also shows responses to these compounds (Stensmyr et al., 2001). Linalool, the main compound in the scent of D. elegans (Table 1), is known to elicit EAD responses in the antennae of several other chafer species (Donaldson et al., 1990; Vuts et al., 2010a, 2010b). Linalool was also found in the floral scents of all seven chafer‐pollinated milkweed species studied by Shuttleworth and Johnson (2010). Based on the efficacy of this compound for attraction of chafer beetles in field trials, Steenhuisen et al. (2013) argued that the shift from bird to chafer beetle pollination in Protea was due to massive upregulation in production of linalool. The role of volatiles in the attraction of hopliine beetles is less clear. Hopliines, including Lepithrix species, are known to respond strongly to color cues (Picker and Midgley, 1996), but there is also evidence that hopliine beetles can be attracted strongly by floral scent (Wragg and Johnson, 2008). Tests of the responses of Trichostetha and Lepithrix beetles to individual volatile compounds are needed to confirm the function of the volatiles emitted by flowers of D. elegans.

We did not observe any consumption of the perianth tissue by beetles in this study. This is notable given that scarab beetles are often associated with “mess and soil” pollination, in which petals and other flower parts are consumed (Bernhardt, 2000; Gottsberger et al., 2012). Trichostetha beetles are considered specialized for feeding on pollen and nectar because of their adentate maxilla (Holm and Marais, 1992). In the case of hopliine scarabs, studies have shown that while some genera are florivorous, others feed almost exclusively on pollen and nectar (Picker and Midgley, 1996). The hopliine genus Lepithrix is considered to belong to the non‐florivorous group and was previously considered to feed only on pollen (Picker and Midgley, 1996), but our observations indicate that Lepithrix beetles feed on nectar from D. elegans flowers (Figure 1E). Interestingly, although both cetonine and hopliine scarabs are known to eat granular pollen of angiosperms (Johnson and Nicolson, 2001), we have found no evidence that these beetles consumed pollinia located in the anther of D. elegans. This is consistent with studies of other orchids pollinated by scarab beetles and indicates that the packaging of orchid pollen into pollinia that are enclosed by anther sacs generally makes it unsuitable for consumption by these beetles.

One approach to the question of whether beetles are effective pollinators of orchids is to consider pollen transfer efficiency (PTE), the fraction of pollen removed by pollinators that reaches stigmas (Johnson et al., 2005; Johnson and Harder, 2023). In this study, we recorded a PTE value of 10.8% in D. elegans, which compares favorably to the mean PTE value of 10% for orchids with massulate pollinaria that was calculated by Harder and Johnson (2008). In the case of epidendroid orchids with solid pollinaria, which have a mean PTE across all pollination systems of ~30% (Harder and Johnson, 2008), an extraordinarily high PTE value of 68% has been reported for the beetle‐pollinated species Orthochilus ruwenzoriensis (Singer and Cocucci, 1997). Peter and Johnson (2014) recorded a PTE value of 25% for a beetle‐pollinated ecotype of the epidendroid orchid Eulophia paviflora, which was notably higher than the value of 6% recorded for the bee‐pollinated ecotype of the same species. There is thus now compelling evidence that beetles can be very effective agents of pollen transfer among orchid flowers. In a recent synthesis, the overall PTE of plants pollinated by beetles was found to be very similar to that of plants pollinated by other insect groups, thus countering the traditional notion that beetles are inefficient pollinators (Faegri and Van Der Pijl, 1979).

The most detailed previous studies of the potential of scarab beetles as agents of pollen dispersal have utilized mark‐recapture methods. Young (1988) reported relatively long movements (83 m; range: 1–529 m) by Cyclocephala dynastine scarabs in a population of Dieffenbachia longispatha Engl & K. Krause (Araceae) in the Neotropics, but flowering individuals were very widely scattered in this population and most beetles were recaptured on nearest neighbors (i.e., an inflorescence nearest to the one on which they were marked during the previous day). In a study of the aroid Xanthosoma daguense Engl., the median movement distances of dynastine scarabs were shorter (33–47 m), reflecting its denser populations (Garcia‐Robledo, 2010). Englund (1993) studied the movement of cetoniine scarabs among Viburnum shrubs in Sweden. As in our study, he found that beetles either spent just a few seconds per plant or spent long periods (≤6 h) on the same plant. The average distance moved was 18.5 m, which exceeded the mean nearest‐neighbor distance of 6 m. Mark‐recapture studies have several limitations. Besides not providing information on within‐plant pollen dispersal, they can overestimate the median pollen dispersal distances because animals may visit several plants before being recaptured, but they could also underestimate these distances if there is extensive pollen carryover or if beetles with pollen regularly fly beyond the search area. Our results from tracking of color‐labeled pollen suggest that pollen of D. elegans is frequently dispersed well beyond nearest neighbors, with some recipients being separated from donors by ≤12 intervening plants (Figure 3). The median pollen dispersal distance of ~6 m recorded for D. elegans in our study is comparable with values reported for populations of other herbaceous plant species (Price and Waser, 1979; Webb and Bawa, 1983; Van Rossum, 2009; Van Rossum et al., 2011). The estimate of 30.4% for the rate of pollinator‐mediated self‐pollination (i.e., geitonogamy and facilitated within‐flower self‐pollination) is comparable to overall values of ~35% obtained for other Disa species (Johnson et al., 2005; Hobbhahn et al., 2017) and rewarding orchids more generally (Peakall and Beattie, 1991; Jersáková and Johnson, 2007), indicating that beetles are comparable to other insect groups as agents of cross‐pollination among flowers.

Overall, our results demonstrate the efficacy of beetles for dispersal of orchid pollen and highlight several floral traits that can be interpreted as adaptations for chafer beetle pollination. Further work should focus on the functional role of volatiles for attraction of the chafer beetles recorded in this study. This could include testing the electrophysiological antennal responses of the beetles to various compounds, including different enantiomers of linalool, as well as field bioassays to confirm the behavioral effectiveness of compounds that elicit antennal responses. The degree of convergent evolution in floral scent chemistry among different plant species that are pollinated by these beetles would also be worth investigating.

Most of the studies of beetle pollination in orchids were published in the past two decades, which is relatively recent in the history of the exploration of orchid pollination systems. Now that older ideas about the absence of beetle pollination among orchids have been firmly dispelled, it seems likely that additional cases will be discovered and described. Traits that seem to be strongly linked with beetle pollination include upward‐facing orientation of flowers, such that they provide a landing platform, strongly contrasting color guide patterns and, in the case of flowers pollinated by large cetoniines, a strong scent dominated by monoterpene alcohols and benzenoid esters. Beetle pollination, as least in South Africa, seems to be found mostly in terrestrial orchids, and this may be linked with the general importance of beetles as pollinators in open grassland and shrubland vegetation in this region (Goldblatt et al., 1998; Johnson and Steiner, 2003; Steenhuisen and Johnson, 2012).

AUTHOR CONTRIBUTIONS

S.D.J.: conceptualization, data acquisition, formal analysis, visualization, chemical analyses, photographic documentation, writing—original draft, review and editing. T.v.d.N.: data acquisition, writing—review and editing. N.H.: data acquisition, formal analysis, writing—review and editing. A.P.: data acquisition, photographic documentation, writing—review and editing.

Supporting information

Appendix S1. Morphological traits of flowering plants of Disa elegans.

Appendix S2. Comparison of the linalool peak in a Disa elegans floral scent sample to retention times of synthentic linalool standards (injections of rac‐(±)‐linalool and R‐(–)‐linalool) on a chiral column.

Johnson, S. D. , Hobbhahn N., van der Niet T., and Pauw A.. 2025. Floral specialization for beetle pollination and its implications for pollen dispersal in an African orchid. American Journal of Botany 112(11): e70117. 10.1002/ajb2.70117

DATA AVAILABILITY STATEMENT

Data available at Zenodo: https://doi.org/10.5281/zenodo.16086736.