Berry Batesian mimicry enables bird dispersal of asexual bulbils in a yam

Abstract

Many plants that abandon sex rely on clonal propagules, but their short dispersal distances can trap offspring near parent plants and enemies. We show that the yam Dioscorea melanophyma—which has lost sexual reproduction—evolved black, glossy bulbils that mimic co-occurring black berries and entice frugivorous birds to ingest and disperse them. Birds from 22 species fed on bulbils, with visits peaking in October–February when true fruits are scarce. Bulbil reflectance overlapped sympatric berries and was indistinguishable in a UV-sensitive avian vision model for many species, consistent with Batesian visual mimicry. Feeding trials with the dominant visitor (Pycnonotus xanthorrhous) showed short gut retention and negligible destruction, so bulbils are excreted largely intact and viable and monitoring revealed that bulbils achieved dispersal distances similar to rewarding endozoochorous plants. Thus, sensory deception that exploits fruit–frugivore signal–reward rules can restore ecologically meaningful movement after the loss of sex in asexual lineages.

Seed dispersal is fundamental for plant reproduction, and it plays key roles in the organization of ecological communities worldwide (1). Yet, some plants have completely lost sexual reproduction, most often via apomixis following hybridization or polyploidy that destabilizes meiosis and embryo-sac development (2, 3). Asexuality can provide reproductive assurance under mate/pollinator limitation and by locking in heterozygous, coadapted gene complexes from hybrids, which facilitates colonization, range-margin persistence, and classic geographical parthenogenesis patterns in which asexual species have larger ranges (4, 5). However, the loss of sexual reproduction also removes recombination, reduces genetic diversity and adaptive potential, and typically restricts propagation to short distances, tying clonal offspring to parents and enemies (2, 4, 5). Thus, plants that have lost sex should be under strong selection to evolve dispersal mechanisms that move propagules beyond parent-proximate hazards, because dispersal governs both the quantity and quality components of recruitment (1, 6).

Clonally reproducing plants can still disperse via gravity (barochory), water (hydrochory), especially along riparian corridors (7), and via animals, when vegetative units attach to animal fur or feather (epizoochory) (8). Endozoochory where vegetative units are ingested and excreted could provide efficient dispersal of clonal units, yet so far, only tiny aquatic Araceae are known to disperse this way (9). However, because the loss of sexual reproduction prevents the rewards through fleshy fruits, asexual propagules cannot “pay” dispersers for their services, thus preventing the use of signal-reward correlations that guides avian foraging (10). These constraints make sensory deception–coopting fruit-like signals without offering rewards—a plausible route to animal transport of vegetative propagules. While deceptive seeds have been found frequently (e.g., refs. 11 and 12), to our knowledge, no deceptive vegetative propagule has been reported so far.

Here, we report the instance of deceptive vegetative propagules to exploit avian endozoochory in the yam Dioscorea melanophyma (Dioscoreaceae). In the field, we identified a jet-black and glossy “berries” (Fig. 1A), that, upon dissection, proved to be seedless bulbils resembling sympatric berries (Fig. 1B). Although the original description of D. melanophyma in 1908 (13) accurately identifies these structures as bulbils, herbarium specimens have often mistaken them for fruits (Dataset S1). This urged us to test whether these black bulbils use deception to lure birds into endozoochorous dispersal. Using camera trapping, reflectance spectroscopy with avian visual modeling, and experiments, we show that D. melanophyma employ visual Batesian mimicry to deceive birds and achieve longer dispersal distance through deceptive endozoochory.

Fig. 1.

D. melanophyma bulbils mimic berries. (A) D. melanophyma bulbils. (B) Sympatric Cipadessa baccifera berries. (C) Ancestral state estimation from stochastic mapping (Materials and Methods) of reproductive systems in Sect. Lasiophyton of Dioscorea. (D and E) Scanning electron microscopy from the bulbil epidermis (D) and pigment-containing epidermal cells (E). [Scale bars are 100 μm in (D) and 2 μm in (E).] (F) Liquid Chromatography-Mass Spectrometry (LC–MS) reveals that the pigment responsible for bulbil color is luteolinidin. (G) Bulbil size range (red) is variable and encompasses that of most sympatric berries. (H) Bulbil spectral curves overlap with those of sympatric berries. Nonmetric Dimensional Scaling (NMDS, Inset) between bulbils and berries spectral curves. (I) In the avian visual space, the JNDs values between bulbils and most berries are significantly less than 1 (red bars).

Results and Discussion

D. melanophyma is a perennial yam vine, distributed in southwestern China (1300 to 2500 m a.s.l), with scattered records from Nepal, India, and Bhutan in the western Himalaya (14). This species solely relies on bulbils, as its sexual reproduction has degenerated and it cannot produce viable seeds. An ancestral state estimation in Dioscorea shows a gradual transition from sexual reproduction, to joint sexual-asexual reproduction, to only asexual reproduction (Fig. 1A). Scanning electron microscopy and chemical analyses revealed that the black color of D. melanophyma bulbils is due to the epidermal cells accumulating high concentrations of the anthocyanin luteolinidin (Fig. 1 D–F, Materials and Methods). To test the extent to which D. melanophyma bulbils mimic berries, we first compared the size and color of these bulbils with those of sympatric berries in 27 plant species (Dataset S2). Measurements showed that D. melanophyma bulbils had the largest size variation, encompassing approximately 80% of the size range of sympatric berries (Fig. 1G and Dataset S1). This range of variation allows D. melanophyma to match berries of various sizes, conferring the potential to attract frugivores of different body sizes.

To assess whether the attractivity of D. melanophyma bulbils resembles that of berries, we measured their reflectance spectra as well as those berries from 27 species with N = 10 replicates per species (Materials and Methods). Spectral analysis revealed a high degree of overlap between the spectral curves of the bulbil surfaces and those of multiple berry species (Fig. 1H). Nonmetric multidimensional scaling (NMDS) further indicated that the sample group of bulbils was well embedded within the sample group of berries (PERMANOVA, F = 1.535, R² = 0.00675, P = 0.186), suggesting that bulbil surface color is close to that of black berries (Fig. 1 H, Inset).

Next, we tested if birds discriminate bulbils from black berries by modeling chromatic contrast (ΔS) as just-noticeable differences in a receptor-noise–limited (RNL) UVS avian model and comparing ΔS to JND = 1 (Materials and Methods). Under the UVS receptor-noise model, ΔS (JND) between bulbils and 15 berry species was significantly <1 (Fig. 1I and Dataset S3), indicating chromatic indistinguishability to birds. For the remaining species, a subset of individuals also had ΔS ≤1, indicating partial overlap. Thus, bulbil surface color closely matches the abundant, bird-favored black berries, consistent with strong potential for avian-mediated dispersal.

We quantified bird–bulbil interactions with arboreal camera traps, defining independent events as image sequences separated by >5 min and classifying them as attraction (approach/inspection) or access (touch/pecking/swallowing/removal). Across N = 27 D. melanophyma individuals monitored over 3 y (2,393 camera-days; 2,994 bulbils in view), we recorded 204 attraction events involving 390 bulbils (attraction rate 13.03%) and 99 access events in which 147 bulbils were removed (removal rate 4.91%) (Fig. 2A and Movie S1). Visitation was strongly seasonal, with 82.35% of events in October–February, when berries are scarce, consistent with a “substitution” effect of the mimic. Removal remained low, as expected for a deceptive display. However, D. melanophyma bulbils remain displayed from July to May –far longer than sympatric berries– extending encounter opportunities during winter scarcity and partly offsetting low per-visit removal rates (Fig. 2B). Activity peaked at 08:00 to 13:00 (61.76%) and 16:00 to 17:00 (23.04%) (Fig. 2C). We documented 22 small passerine species; Pycnonotus xanthorrhous was dominant (83 attraction events; relative abundance index [RAI], 40.69/204; removal RAI 48.48), with Actinodura cyanouroptera, Copsychus saularis, Phoenicurus auroreus, Pterorhinus sannio, Pycnonotus aurigaster, and Schoeniparus dubius collectively contributing RAI 45.10, and the remaining 15 species accounted for RAI 14.22 (Fig. 2E).

Fig. 2.

Endozoochorous dispersal of bulbils substantially increases the dispersal potential of D. melanophyma. (A) Brown-breasted bulbul feeding on a D. melanophyma bulbil. Monthly (B) and daily (C) patterns of attracting events to bulbils. Blue lines show the frequency of attraction events when a bird removed a bulbils and orange lines correspond to successful ingestions. (D) Choice experiments with N = 4 brown-breasted bulbuls in which D. melanophyma bulbils vs. Solanum nigrum berries were offered to the birds and either unsupplemented, or continuously supplemented to maintain sample sizes. (E) Relative abundance index (RAI) of bird diversity, showing the total (attraction to bulbils) and effective (successful bulbil ingestion) RAIs. (F) Radial dispersal kernel for D. melanophyma (Materials and Methods). Solid: baseline (median D). Shaded: 95% D-sensitivity. Dashed: median, 90th, 95th, 99th percentiles. (G) Tail probability (survival) S(r) = Pr(R > r) on log scale. Solid: baseline; thin lines: low/high D bounds. Dots: S(250 m), S(500 m), S(1,000 m). See also Movie S1.

To test whether birds prefer berries to deceptive bulbils, and whether they can discriminate them, we conducted cafeteria experiments with N = 4 brown-breasted bulbuls where we offered N = 10 bulbils and N = 10 Solanum nigrum berries (Materials and Methods). In a first experiment, we replaced each berry or bulbil as it was consumed to maintain the same sample sizes throughout. This showed that birds preferred rewarding berries, yet they significantly consumed bulbils too (Mann–Whitney U test, U = 0.0, Z = −2.67, P < 0.05) (Fig. 2D). Birds could not visually discriminate bulbils from berries, but dropped them more frequently (Mann–Whitney U test, U = 0.0, Z = -2.80, P < 0.05). In a second experiment, using the same set-up but without replacement to mimic winter berry rarefaction, there were no significant differences between berries and bulbils consumption (Mann–Whitney U test, U = 10.5, Z = −0.41, P = 0.68) (Fig. 2D). Altogether, these data show that although overall removal is low, D. melanophyma bulbils are eaten by birds, and successful removal becomes more frequent as berries become rarer.

We next tested whether bulbils can survive passage through the avian digestive tract. Brown-breasted bulbuls, the primary bulbil feeders, were fed bulbils and their egesta collected (Materials and Methods). Egested bulbils were planted in situ at KM under the same conditions as noningested controls, and we monitored their germination and decay over the following year. The results showed that the average gut-retention time of the bulbils was 24.30 ± 8.94 min (N = 168 timed egesta); of the 314 fed bulbils, only three were not recovered (0.96% digestion), indicating near-universal intact passage. Growth monitoring revealed no significant difference in germination rate between the egested (N = 77) and the control bulbils (N = 112) (86.50% vs. 92.06%; chi-square test, χ² = 1.27, P = 0.262).

Next, we estimated movement from postfeeding fly-outs and the Gamma gut-retention time to obtain a mechanistic radial kernel (Fig. 2F; Materials and Methods). The kernel predicts a median dispersal of ~230 m (95th ≈ 0.52 km, 99th ≈ 0.67 km) (Fig. 2F). Tail probabilities show that ~6% of events exceed 500 m and ~0.36% exceed 750 m, quantifying rare but consequential long-distance transport (Fig. 2G; ref. 6).

This study identifies Batesian mimicry in an asexual reproductive structure, revealing that bulbils attract birds by mimicking the visual signals of sympatric berries in morphology and color. This strictly asexual yam co-opts berry–bird networks through visual Batesian mimicry: Black, luteolinidin-colored bulbils are perceived as fruits, are occasionally swallowed and egested intact, and thus achieve deceptive endozoochory. These results demonstrate a viable route by which loss of sex –and its attendant dispersal bottlenecks– can be offset by sensory deception that exploits signal–reward correlation in frugivory, with rare but consequential transport events extending propagule shadows, potentially further increased by secondary dispersal by predatory birds (15). Thus, deceptive signaling can restore ecologically meaningful movement to asexual lineages.

Materials and Methods

From 2019 to 2025 we surveyed D. melanophyma in China and Nepal and ran experiments on a Kunming population. We inferred reproductive-mode evolution on a Dioscorea phylogeny and identified bulbil pigment using cryo-SEM and UHPLC-HRESI-MS. Berry mimicry was quantified from bulbil/berry size, reflectance (300 to 700 nm), and avian discriminability modeled with a UV-sensitive receptor-noise–limited JND framework. Bird interactions were recorded year-round with arboreal camera traps, preferences tested in two-choice cafeteria assays, endozoochory assessed via gut-passage/viability trials, and dispersal estimated from fly-out displacement with a Gamma gut-retention distribution. Full methods are provided in the SI Appendix, Supplementary Materials and Methods.

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (XLSX)

Dataset S02 (XLSX)

Dataset S03 (XLSX)

Dataset S04 (XLSX)

Dataset S05 (R)

Movie S1.

Images from the tracking camera showing birds ingesting D. melanophyma bulbils, feeding it to their young, felling after being removed, and flying away with a bulbil.

Data, Materials, and Software Availability

Study data are included in the article and/or supporting information.