Autogamy vs. Allogamy: Biotic-Dependent Populations Reveal Shift in the Reproduction System of an Essentially Autogamous Vanilla

Abstract

Flowering plants have evolved a wide range of strategies to avoid self-fertilization. Although allogamy is common among members of the genus Vanilla, V. lindmaniana is widely known for its ability to self-fertilize, resulting in fruit-set rates of 83–100% following self-pollination. During studies of Vanilla species in the Brazilian Cerrado, populations of V. lindmaniana exhibiting low fruit set under natural conditions were identified. Here, the factors associated with low fruit set in these Cerrado populations were investigated through analyses of floral traits, field observations of pollinators, and experimental treatments designed to assess the breeding system. The flowers of V. lindmaniana are adapted for bird pollination. Although the species commonly produces fruits through facultative autogamy, flowers from the Cerrado populations exhibited a stigma that was separated from the anther by a large rostellum, preventing self-pollination. While autogamy is common in V. lindmaniana, completely allogamous populations may also occur. Such a mixed breeding system may provide advantages over reproductive systems based exclusively on autogamy. Furthermore, it may promote reproductive assurance in the event of pollinator decline or in environments where biotic pollination is irregular or absent.

1. Introduction

Plants are essentially sessile organisms that can expand their distribution range through seed dispersal and transfer their genes through pollination [1]. Flowering plants have evolved a wide range of strategies, including pre- and post-zygotic barriers, to avoid self-fertilization. Post-zygotic barriers act during fertilization or embryonic development, whereas pre-zygotic barriers prevent pollen transfer between the anther and the stigmatic surface (e.g., mechanical and temporal barriers) or inhibit pollen tube development (e.g., genetic barriers) [2]. Although genetic barriers, such as sporophytic and gametophytic self-incompatibility, have been reported in Orchidaceae, the floral mechanisms preventing self-pollination are particularly diverse within this large plant family [3]. The diversity of mechanical barriers in orchids is remarkable. However, the principal floral structure preventing contact between the stigma and the anther in Orchidaceae is the rostellum [4,5]. In autogamous orchids, the rostellum is often reduced or absent, allowing contact between the stigmatic surface and self-pollen [5]. With the exception of species bearing cleistogamous flowers, in which self-pollination is obligatory, autogamous orchids with chasmogamous flowers generally exhibit facultative self-pollination. Under this reproductive strategy, flowers open and become available to pollinators; however, if pollination by biotic vectors does not occur, fruits are produced through autonomous self-pollination (autogamy). Facultative self-pollination has been documented in numerous orchid species, including members of the genus Vanilla [6,7].

The genus Vanilla comprises more than 100 species distributed throughout the tropical regions of Africa, Asia, and the Neotropics [7,8,9]. The Neotropics represent the center of diversity for the genus, harboring approximately 60 recognized species. With more than 40 species and one nothospecies, Brazil is considered the main center of diversity for Vanilla, including numerous endemic taxa [7,8,9,10,11]. Species of Vanilla are predominantly melittophilous [12]. However, at least one Neotropical species, Vanilla lindmaniana, exhibits a pollination system in which allogamous pollen transfer is mediated by birds [6,12]. Vanilla lindmaniana offers nectar as a floral reward and is pollinated by hummingbirds (Trochilinae). Although cross-pollination is mediated by birds, the species also exhibits facultative self-pollination [6,7]. Despite possessing a well-developed rostellum, pollen grains can come into contact with the margins of the stigmatic surface, thereby enabling self-pollination [6]. As a consequence, natural fruit set in V. lindmaniana is typically high, with 83–100% of flowers developing into fruits under natural conditions [6,7]. Although high fruit production through self-pollination is considered a common feature of V. lindmaniana [7,12], populations exhibiting low fruit set under natural conditions were identified during studies of Neotropical Vanilla. Because reduced fruit set has never previously been reported for this species, this study aimed to investigate the following: (1) whether barriers preventing fruit formation through autogamy occur in populations from the Brazilian Cerrado; (2) whether biotic pollination occurs, as previously reported for Amazonian populations [6]; (3) whether floral morphology and floral resources are associated with the pollination system, and the identity of the pollinators; (4) whether pollination in V. lindmaniana exhibits any degree of specialization; and (5) whether variations in breeding systems occur across the species’ distribution range.

2. Materials and Methods

2.1. Study Sites and Plant Material

The study of the reproduction system of Vanilla lindmaniana was based on populations occurring in the boundaries of the municipalities of Firminópolis (16°34′55″ S, 50°18′18″ W; 672 m a.s.l), state of Goiás, Central-western Brazil, and Manaus (3°4′25″ S, 60°0′20″ W; 39 m a.s.l), state of Amazonas, Northern Brazil (Figure 1). The climate of Manaus is humid tropical (‘Afi’) according to Köppen’s classification [13]. The temperature varies from 24 to 27 °C and is rarely below 18 °C. Rainfall occurs throughout the year [14]. Amazonas is located in the Amazonian Biome. The climate of Firminópolis is defined as “Aw” (Tropical savanna, with a well-defined wet season) [13]. The average annual temperature is ca 23 °C. The climate is seasonal, with a rainy and hot summer (December to March) and a dry and milder winter (May to September). Firminópolis is located in the Cerrado Biome [15]. In Manaus, the plants were found growing on the swamp palm “buriti” (Mauritia flexuosa Linnaeus), while in Firminópolis, V. lindmaniana grows on the “bacuri” palm (Attalea phalerata Mart. ex Spreng.).

To study details of the floral biology and breeding system, 14 stem cuttings (ca. 1 m) were collected from plants of both studied regions, with seven of them from a population growing in the municipality of Manaus, and five from a population growing in the municipality of Firminópolis. Each stem cutting was collected from a different phorophyte. The phorophytes were at least 100 m apart. The collected cuttings of V. lindmaniana were planted individually in ceramic pots with a mixture of pine bark, Sphagnum, and litterfall. Plants were cultivated in the LBMBP Orchid House, University of São Paulo (FFCLRP-USP), municipality of Ribeirão Preto (approx. 21°10′ S, 47°48′ W; 546 m a.s.l.). Ribeirão Preto is located in an ecotone that is characterized as being between the Atlantic Forest and the Cerrado [16]. The LBMBP Orchid House is completely covered with 50% black shade cloth in order to prevent any contact between the plants used in experiments with flower visitors. Fieldwork and specimen collection were conducted under a permit issued by the Brazilian Ministry of the Environment through the SISBIO system (Permit No. 35178-1), granted by the Instituto Chico Mendes de Conservação da Biodiversidade (ICMBio).

2.2. Floral Features

The morphological features of 30 fresh flowers (10 plants; 10 inflorescences; 3 flowers per plant) were analyzed under a binocular stereomicroscope. Floral details were studied and photographed with a Stereozoom Leica S8 APO stereomicroscope (Leica Microsystems, Wetzlar, Germany) attached to a PC using M50 image analysis software version 1.0 and a Nikon D-SLR D800 (Nikon Corporation, Shinagawa, Tokyo, Japan) camera equipped with a Micro Nikkor 105 mm f2.8 lens.

The flowers used for morphological study were examined for the presence of exudates [17]. A histochemical investigation with Fehling’s reagent was performed to identify the presence of reducing sugars in the nectar [18]. Appropriate controls (slides with no coloration) were run simultaneously with the histochemical test, a methodology similar to the one used in the study of secretory structures of Neotropical Vanilla [19]. Slides were examined under a Leica DM500 (Leica Microsystems, Wetzlar, Germany) light microscope, and the images were captured with a Leica ICC50 (Leica Microsystems, Wetzlar, Germany) video camera attached to a PC using IM50 image analysis software version 1.0.

The nectar volume and concentration of V. lindmaniana flowers were measured with a 50 µL Hamilton syringe and a Bellingham and Stanley (series Eclipse) hand-held refractometer, respectively [20]. Nectar measurements were made on thirty 1-day flowers (10 inflorescences; 10 plants). The measurements were taken in the morning hours, between 0800 and 0900 h.

2.3. Floral Visitors and Pollination Process

Focal observations of visitors on flowers of V. lindmaniana were carried out between 15–22 November 2023 and 20–27 November 2025. Each daily period of focal observations lasted from 08:00 to 17:00 h, totaling 128 h. Observations were conducted on the surroundings of the LBMBP Orchid House, Ribeirão Preto, Campus of the University of São Paulo, on four plants bearing more than 20 flowers (three inflorescences) each. Visitors to flowers were recorded using a Nikon D-SLR D800 (Nikon Corporation, Shinagawa, Tokyo, Japan) camera and a Micro Nikkor 105 mm f2.8 lens, as well as a Sony FDR-AX100 4k Handycam (Sony, Tokyo, Japan) attached to a 12 V 60 Ah battery and a 200 W power inverter 12 V DC to 127 V AC. Visitors were identified with the help of a field guide regarding the birds of Brazil [21]. To assess whether birds were effective pollinators of V. lindmaniana, pollen grains were collected from the beaks of hummingbirds and compared with pollen from flowers of V. lindmaniana cultivated at the LBMBP Orchidarium.

2.4. Breeding System

The experimental treatments for the investigation of the breeding system of Vanilla lindmaniana were performed in the LBMBP Orchid House. A total of 10 plants (5 per study site) were used in the experimental treatments. The following treatments were conducted using intact flowers: manual self-pollination to test the occurrence of self-compatibility, autonomous self-pollination to test the incidence of autogamy, emasculation to test the occurrence of apomixis, and cross-pollination. A total of 180 flowers (90 flowers per study site; 30 flowers per treatment) were used. All treatments were randomly applied to each inflorescence using one-day flowers. Treatments were performed during the morning hours, between 0800 and 0900 h. An entire anther was used in each experimental pollination event (manual self- and cross-pollination). Cross-pollination was performed with previously emasculated flowers and by using the pollen of a plant from a distinct pot. To evaluate whether the number of flowers setting fruit is influenced by pollination limitations in each experimental treatment, a GLM analysis with a binomial distribution and a logit link function was performed. Model significance was evaluated using Wald statistics, and goodness-of-fit was assessed through residual analyses. All analyses were conducted in R version 4.5.3 using the glm() function in the stats package. The fruit sets in successful treatments of manual self-pollination, cross-pollination, and autonomous self-pollination involving plants from both populations studied were compared using a t-test for independent samples with Statistica 6.0 [22]. Statistics with relevant values are presented.

The fruit sets of 20 plants under natural conditions (open pollination) (10 individuals per study site, one inflorescence per plant) were recorded. The fruit set between both study areas was compared using a t-test for independent samples with the software Statistica 6.0 [22].

3. Results

3.1. Floral Features

Plants of Vanilla lindmaniana from the studied populations produced axillary inflorescences bearing up to 33 (Firminópolis) and 24 (Manaus) flowers, which opened successively. A single yellow flower opened each morning and remained functional for approximately 9–10 h. The labellum exhibited a central cavity extending to the entrance of the nectar chamber. The anther was versatile, and the pollen monads were aggregated into a pollinium-like mass. In plants from the Firminópolis population, the stigmatic surface was separated from the anther by the rostellar flap (Figure 2A–C). In contrast, in flowers from the Manaus population, the pollen mass was not clearly separated from the stigma because the rostellum was narrower than the anther (Figure 2D–F). In the Firminópolis population, the anther measured 4.2–4.3 × 3.1–3.2 mm, whereas the pollen mass and rostellum measured 2.7–2.8 × 2.6–2.7 mm and 3.2–3.3 × 3.3–3.4 mm, respectively (Figure 2A–C). In the Manaus population, the anther measured 4.1–4.2 × 3.2–3.3 mm, whereas the pollen mass and rostellum measured 2.7–2.8 × 2.7–2.8 mm and 3.2–3.3 × 2.5–2.6 mm, respectively (Figure 2D–F). The ventral portion of the column bore white trichomes near the stigmatic surface (Figure 2A–F and Figure 3A). The yellow labellum exhibited a central cavity that extended to the entrance of the nectar chamber (Figure 3A). The margins of the labellum were adnate to both sides of the column from the base to the upper third, resulting in a tubular flower. The lower third of the floral tube formed a nectary approximately 20 × 2 mm in size (Figure 3A). The lower half of the nectary was smooth, whereas the upper half was lined with hyaline trichomes (Figure 3A).

The secretory tissue of V. lindmaniana flowers was located in the inner third of the labellum (Figure 3A). It consisted of a cluster of homogeneous, loosely arranged cells (Figure 3A). Nectar accumulated within the nectar chamber (Figure 3A,B). The sugary secretion reacted positively to the Keto-Diabur-Test^® (Roche; Figure 3C). Fehling’s test was also positive for reducing sugars, as indicated by the formation of dark brown precipitates (Figure 3B). The nectar volume ranged from 2 to 8 μL (mean = 4.9 ± 1.3 μL), whereas the sugar concentration ranged from 11% to 23% (mean = 16.7 ± 3.5%).

3.2. Floral Visitors and Pollination Process

Vanilla lindmaniana was visited by two bird species, Eupetomena macroura (Gmelin, 1788) (Figure 4A,B; Supplementary Material, Video S1) and Coereba flaveola (Linnaeus, 1758) (Supplementary Material, Video S1). However, the only effective pollinator recorded in the study area was the hummingbird E. macroura (Figure 4A,B; Supplementary Material, Video S1). Pollinators visited the single flower available per inflorescence, and each visit lasted between 4 and 11 s. Visits occurred throughout the day. During the observation period, 78 visits by E. macroura were recorded. Visits began with the hummingbird hovering in front of an inflorescence before inserting its bill into the floral tube (Figure 4A). The central cavity of the labellum guided the bill toward the nectar chamber, where floral nectar was stored (Figure 4A). Pollen removal occurred when the hummingbird contacted the anther with its bill while probing for nectar (Figure 4B). In contrast, the passerine C. flaveola was considered merely a floral visitor because its visits did not result in pollen removal (Supplementary Material, Video S1). In addition to the 78 legitimate visits, E. macroura was recorded piercing the base of the labellum to access nectar 41 times. These visits were therefore classified as nectar robbing events (Supplementary Material, Video S1). The pollen collected from the beaks of hummingbirds was morphologically identical to that of V. lindmaniana.

3.3. Breeding System

Plants from both populations of V. lindmaniana were fully self-compatible. Fruit set following manual self-pollination reached 96.6% in Firminópolis and 100% in Manaus, whereas fruit set following cross-pollination reached 93.3% and 96.6%, respectively (

Table 1. Percentage of fruit set of Vanilla lindmaniana according to the experimental treatments and under natural conditions (open pollination) in plants from two study sites: Firminópolis, Central-western Brazil (a), and Manaus, Northern Brazil (b). Figures in parentheses indicate the number of flowers/fruits.

Treatment	Fruit Set (a)	Fruit Set (b)
Autonomous self-pollination	(30/0) 0%	(30/28) 93.3%
Manual self-pollination	(30/29) 96.6%	(30/30) 100%
Manual cross-pollination	(30/28) 93.3%	(30/29) 96.6%
Open pollination	(450/90) 20.7%	(485/458) 94.4%

). The autonomous self-pollination treatment resulted in a fruit set of 93.3% in the Manaus population, whereas no fruits resulting from autogamy were recorded in the Firminópolis population (Table 1). In flowers from Manaus, pollen grains naturally contacted the stigmatic surface, resulting in autonomous self-pollination and fruit production. In contrast, the large rostellum present in flowers from Firminópolis prevented contact between the pollen mass and the stigma. Consequently, pollen transfer depended on a biotic pollination vector. No significant differences in fruit set were detected between the autonomous self-pollination and manual self-pollination treatments in the Manaus population (t = −1.00, df = 29, p = 0.32), nor between the cross-pollination treatments performed in the Firminópolis and Manaus populations (t = −0.58, df = 52, p = 0.56). The GLM analysis revealed no significant effect of flower number on the fruit set in the autonomous self-pollination treatment involving plants from Manaus (β = −0.029 ± 0.053, z = −0.551, p = 0.582). These plants exhibited a high fruit set, indicating high reproductive efficiency and low pollination limitation. The GLM analysis revealed no significant effect of flower number on the fruit set in plants from the Firminópolis population (β = −0.021 ± 0.019, z = −1.10, p = 0.27). These plants exhibited a lower fruit set, indicating reduced reproductive efficiency and stronger limitation by pollen transfer.

Under natural conditions, the fruit set was 94.4% in Manaus and 20.7% in Firminópolis (Table 1). Fruit sets differed significantly between the two populations (t = −16.91, df = 45.35, p < 0.001).

4. Discussion

The floral features of Vanilla lindmaniana observed in the studied population are largely consistent with those reported from other regions throughout the species’ distribution [6,7]. Vanilla lindmaniana produces short-lived, resupinate flowers that open successively along lateral racemes. Whereas most Neotropical Vanilla species bear whitish or greenish flowers, the production of entirely yellow perianth segments, as observed in V. lindmaniana, is uncommon among its Neotropical congeners. Aside from V. lindmaniana, V. palmarum, V. pompona, V. cribbiana, and the natural hybrid V. × robusta have been reported to possess completely yellowish flowers [23]. The yellow-flowered species V. lindmaniana and V. palmarum form a clade within a basal lineage of the Neotropical group, characterized by non-membranous leaves [7,24]. Among thick-leaved Vanilla species, V. cribbiana and V. pompona belong to the euglossinophilous clade and are pollinated by males of Eulaema [12,19,25,26]. In this clade, pollination is associated with fragrance collection from floral tissues and nectar foraging within the labellum [17,26]. Members of the euglossinophilous clade typically possess a penicillate callus inside the floral tube, whose function is to elevate the bee’s body and facilitate contact between the scutellum and the reproductive structures of the flower [12,19,26]. In contrast, flowers of V. lindmaniana lack a penicillate callus. Instead, they possess a longitudinal cavity within the labellum that guides the hummingbird’s bill toward the nectar chamber [6]. In addition to this specialized floral morphology and the provision of nectar as a reward, V. lindmaniana exhibits other traits commonly associated with ornithophily, including scentless flowers and brightly colored yellow perianth segments. The absence of floral fragrance in V. lindmaniana likely represents a derived condition, as this bird-pollinated species probably evolved from ancestors pollinated by hymenopterans [6,17,25,26,27]. Indeed, gains and losses of floral traits are frequently driven by pollinator-mediated selection [28]. In bird-pollinated plants, such selective pressures often result in scentless flowers, conspicuous flower coloration, and highly specialized floral morphologies [29,30]. Bird pollination is rare in Vanilla, as most species within this pantropical genus are adapted for pollination by hymenopterans [17,26]. Although hummingbird visitation has been reported in V. planifolia [27] and V. chamissonis, both species exhibit floral traits associated with melittophily and belong to the euglossinophilous clade [19,24]. While hummingbird pollination in V. lindmaniana had been previously documented, direct evidence of pollen deposition on the beaks of pollinators was lacking because observations were conducted in palm canopies more than 10 m above the ground [6]. The present study provides clear evidence of pollen deposition on hummingbird bills, reinforcing the role of birds as effective pollinators of this species.

Although most Vanilla species depend on biotic pollination for fruit production, autogamy has evolved independently in several lineages, suggesting convergent evolution of breeding systems within the genus [6]. Autonomous self-pollination appears to be particularly widespread among the most basal lineages of thick-leaved Vanilla. Indeed, autogamy has been reported in V. bicolor [31], V. savannarum (= V. lindmaniana?) [7,32], V. palmarum [7,33], and V. lindmaniana [6,7]. Until now, V. lindmaniana was considered a facultatively autogamous species in which autonomous self-pollination coexists with biotic pollen transfer, resulting in both self- and cross-fertilization [6]. However, the present study demonstrates geographic variations in the breeding system across the species’ distribution. Whereas facultative autogamy appears to be the predominant condition in V. lindmaniana [6,7], entirely allogamous populations may also occur. To our knowledge, this study provides the first evidence of obligate dependence on cross-pollination in populations of this widespread species. A mixed breeding system, such as that observed in V. lindmaniana, may offer advantages over reproductive systems based exclusively on autogamy because strictly selfing populations often exhibit high levels of homozygosity [34]. Furthermore, facultative selfing may provide reproductive assurance under conditions of pollinator scarcity or in habitats where pollinator service is unreliable [35,36]. Variations in breeding systems are common among widely distributed plant species, particularly those that have undergone island colonization from continental source populations [37]. This may be relevant to V. lindmaniana, whose colonization of Caribbean islands appears to have been mediated by migratory birds originating from Amazonian populations [38]. While facultative selfing can ensure reproduction in the absence of pollinators, biotic cross-pollination in stable continental populations is expected to increase genetic diversity and, consequently, adaptive potential relative to strictly autogamous lineages [3,37,39]. In the Cerrado populations studied here, the requirement for bird-mediated pollen transfer is imposed by a pre-pollination barrier, namely the enlarged rostellum, which prevents contact between the pollen mass and the stigma and thereby promotes outcrossing. Although cross-pollination clearly provides evolutionary advantages over strict selfing, autogamy and genetic self-compatibility may represent desirable traits in economically important species. The vanilla flavor is among the most valuable spice products in the world, and commercial production remains heavily dependent on labor-intensive manual pollination. Therefore, knowledge of reproductive strategies, breeding systems, and the natural history of wild Vanilla species is essential for identifying favorable genotypes and developing improved breeding programs for vanilla cultivation [12].