Bloodstained flowers and bloodthirsty flies

Flowers of C. gerrardii are unusual even by the kaleidoscopic standards of South Africa's flora, as they resemble velvet-green, whiskered starfish with a coat of morning dew (see fig. 1a in Heiduk et al., 2023). Who could pollinate such a flower, and how would they find it? The authors answer these questions with a diversified tool kit, including chemical analyses of the floral scent and dew-like secretions, spectrometric and electron-microscopic explorations of flower color and surface texture, field observations of pollinators, and electrophysiological and behavioral assays measuring their responses to the floral bouquet. Their findings are unexpected and provocative.

Flowers of C. gerrardii are pollinated by minute ‘jackal flies’ (Desmometopa spp., Milichiidae), called ‘kleptoparasites’ because they steal the prey of spiders or mantids by drinking their blood (hemolymph). Previously, Heiduk et al. (2015, 2016) identified similar flies as pollinators for other species of Ceropegia, using the term ‘kleptomyiophily’ (literally, ‘lover of thieving flies’), a term coined by Oelschlägel et al. (2015) in a similar study, to describe pollination by kleptoparasitic flies. In these cases, the flowers formed tubular chambers that entrapped the flies, compensating for inefficient pollen transfer by extending the flies' residence time. By contrast, the open flowers of C. gerrardii detain flies by secreting liquid globules containing sugar and protein, a substance closer in composition to insect hemolymph than to floral nectar. This finding recalls earlier research on seed dispersal mutualisms mediated by elaiosomes, the food bodies attached to ant-dispersed seeds in temperate forest herbs. The nutritional content of elaiosomes, including free fatty acids, amino acids, and the disaccharide trehalose, is more similar to that of prey fed to ant larvae (again, insect hemolymph) than the seeds to which they are attached (Fischer et al., 2008). Similarly, C. gerrardii plants enlist jackal flies as pollinators by providing a floral reward that mimics their primary source of nutrition: the spilled blood of bees.

How are such pollinators attracted? Jackal flies arrive rapidly at a kill, hunting wounded insects by responding to cues of their distress (Heiduk et al., 2015). Indeed, two of the Desmometopa fly species that pollinate C. gerrardii arrived within 15 s when wounded honey bees were presented in a natural setting. Hence, the floral scent of C. gerrardii should: (1) mimic the chemical signature of injured honey bees; and (2) serve as a key attractant for jackal flies. Chemical analysis confirmed that the volatile cocktail includes components of honey bee alarm pheromone (isoamyl acetate), Nasonov gland (geraniol), and mandibular gland secretions (2-heptanone; Heiduk et al., 2016). However, an additional 140 volatiles were present in the blend, complicating the task of identifying key attractants. To this end, the authors employed bioassay-guided fractionation, a methodological mantra in the field of chemical ecology (Murphy & Feeny, 2006). First, they established positive controls (wild fly attraction to wounded bees and to floral extracts in acetone), and then they tested artificial blends (fractions) of C. gerrardii scent containing subsets of the volatile bouquets that had triggered electrophysiological responses from jackal fly antennae. Fly responses to different fractions revealed a necessary and sufficient blend of nonan-2-ol, heptan-2-one, geraniol and octyl acetate, the olfactory essence of a ‘bleeding bee’.

What of the ‘green-on-green’ visual display of C. gerrardii flowers? The malachite-colored corolla lobes, when modeled within the color-perceptual space of flies, would not stand out from background vegetation, suggesting that pollinator attraction is scent-driven. The velvety epidermal surfaces include dome-shaped cells and glandular trichomes likely responsible for the hemolymph-like secretions consumed by visiting flies. Thus, specialized pollination in this system results from being less conspicuous to generalized visitors, targeting specific pollinators with chemical mimicry of distressed bees, and presenting a floral reward that nutritionally approximates bee blood. In this vein, it is worth testing whether healthy honey bees are repelled by flowers of C. gerrardii. It remains unclear whether the vibratile hairs at the base of each corolla lobe serve as rails guiding flies toward the floral sex organs or whether they visually mimic the presence of additional flies, a masquerade common to other fly-pollinated plants (Ren et al., 2023). More analyses are needed to determine how closely the sugar and protein contents of the corolla lobe secretions match those of honey bee hemolymph, and thus, the extent to which these floral secretions represent nutritious meals or junk food for jackal flies.

Evolutionary innovations are most convincing when independent examples can be identified from geographically distinct locales. Scent-mediated kleptomyiophily has evolved in parallel in Ceropegia species with chamber-trap flowers, with identified scent components that mimic wasp venom (C. dolichophylla; Heiduk et al., 2015, in China) or (once again) honey bee hemolymph (C. sandersonii; Heiduk et al., 2016, in Southern Africa). The frit fly family (Chloropidae) includes kleptoparasitic species whose females drink insect hemolymph or vertebrate eye secretions. Frit flies also visit Ceropegia flowers and are pollinators of a Mediterranean pipevine (Aristolochia rotunda) whose scent mimics the pungent hemolymph of mirid bugs (Oelschlägel et al., 2015). Across the planet, frit flies also pollinate Australian Corunastylis orchids, which appear to reward them with tear-like secretions from special floral organs (Ren et al., 2023).

Taking a broader view of floral mimicry, this study reveals multiple examples of cognitive misclassification. First, jackal flies misclassify the flowers of C. gerrardii as wounded, bleeding honey bees, consuming their floral secretions as surrogate hemolymph. Second, pollination biologists have misclassified many Ceropegia flowers as brood-site mimics, given that hundreds of species have chamber trap flowers like those of dung- or carrion-mimicking aroids (Arum) and pipevines (Aristolochia). Heiduk and colleagues remind us that kleptomyiophilous flowers are food-deceptive (in C. sandersonii and C. dolichophylla) or rewarding (in C. gerrardii) rather than brood-site deceptive, attracting female flies that consume hemolymph rather than laying eggs. Third, taxonomists have misclassified their concept of the genus Ceropegia, misled by rampant homoplasy (repeated evolutionary gain and loss) in growth form and floral function. Recent phylogenetic evidence supports a more inclusive Ceropegia in which several clades of succulent milkweeds (Hoodia, Stapelia) and Brachystelma are nested, swelling its ranks to 717 species in 63 sections (Bruyns et al., 2017). This semantic change has little ecological impact as, to paraphrase Juliet Capulet; ‘a Stapelia by any other name ([aside], a Ceropegia) would smell as fetid’. However, a more profound evolutionary impact is the realization that ‘stinking starfish’ flowers have evolved several times from ancestors with tubular corollas (Ollerton et al., 2017), including well-known carrion mimics and the bizarre, bleeding flowers of C. gerrardii. These examples underscore the potency of the South African flora as an engine of floral diversification.