Introduction

Approximately 80 million years ago, the Vanilloideae subfamily diverged within Orchidaceae1, which includes the genus Vanilla Miller. This genus of culinary and cosmetic value includes around 140 species2. Among these, only Vanilla planifolia Andrews is cultivated on a large scale, with Vanilla pompona Schiede and the hybrid Vanilla × tahitensis (V. planifolia Andrews ~ Vanilla odorata C.Presl) cultivated to a lesser extent3. They are cultivated for their fruits rich in the organic compound vanillin4,5. Vanilla planifolia crops typically have low genetic diversity due to the use of a small number of clones for reproduction6,7, which can result in low resilience to environmental changes and biotic factors such as pests and pathogens8. Moreover, modern plantations rely on cultivated clones originally introduced from Mesoamerica to the rest of the world during the late 16th century9. Certain varieties, such as the high yielding ‘Mansa’ variety (with more than 75% of fruits retained), are commonly marketed for establishing new plantations, resulting in an industry that relies on a limited genetic base10.

Although commercially V. planifolia has been the most exploited species, the neotropics have around 41 aromatic Vanilla species with the potential for commercial use11. In Costa Rica, two wild species with fragrant fruits, V. odorata and V. pompona, represent critical genetic resources for Vanilla cultivation and could potentially contribute to the vanilla extract trade2, despite being classified as endangered by the International Union for Conservation of Nature (IUCN). Since 2004, studies on genetic diversity in some Vanilla species have been conducted using markers ranging from RAPDs to SNPs. Most of these studies have focused on species-level diversity, with few examining population-level genetic diversity and structure3,6,8,9,12,13,14,15,16,17. In these studies, V. dilloniana, V. imperialis, and V. pompona exhibited higher average heterozygosity compared to V. planifolia, V. odorata, and V. × tahitensis5,9. Despite efforts to advance our understanding of Vanilla genetics, a significant research gap remains concerning the clonal structure, population genetic diversity, and genetic structure of wild V. odorata and V. pompona populations. This knowledge gap, essential for effectively managing and conserving these species, will be addressed in this study.

Species with allogamous reproduction and the ability for long-distance seed dispersal are likely to exhibit higher levels of genetic variation and lower levels of population differentiation compared to species that regularly self-fertilize and have limited dispersal of propagules18. While certain species within Vanilla reproduce naturally via allogamous sexual reproduction or self-pollination, the overall success of natural pollination in Vanilla is low, as commonly observed in other food-deceptive orchids. For example, natural fruit set in V. planifolia in Puerto Rico and Central America is between 1 and 3%19, while Soto Arenas (1999) reported an even lower rate in Mexico: 1 fruit per 100–1000 flowers22. This is consistent with the 0.65% fertilization ratio reported by Pemberton et al. (2023) in southern Florida23 and the 4.9% found by Quezada-Euán et al. (2024) in Yucatán, Mexico24. Therefore, vegetative or clonal propagation is the most plausible and economically viable type of reproduction in the genus25. This may result in widespread spatial distribution of single genets. In natural environments, Vanilla individuals may expand over large areas; for example, an individual of V. planifolia was observed to cover areas up to 0.2 hectares22.

Although populations of V. odorata and V. pompona have been documented in Costa Rica, their genetic structure and diversity remain unknown. The primary objective of this project is to study the genetic diversity and clonal structure of V. pompona and V. odorata populations in Costa Rica using microsatellites. We hypothesize that V. pompona exhibits higher genetic diversity than V. odorata due to its wider distribution and larger population sizes. However, because V. odorata and V. pompona can reproduce clonally, we expect low overall genetic diversity, significant fine-scale genetic structure, and significant differences in allele frequencies among populations. Understanding the genetic diversity of these Vanilla species will provide baseline information for conservation strategies, such as increasing population connectivity or providing guidelines on best practices for translocating individuals among populations. It will also encourage the use of these species to improve the genetic resources of extensively cultivated Vanilla. A comprehensive understanding of the genetic diversity and reproductive biology of Vanilla species, including clonality and inbreeding in fragmented and geographically distant populations, is essential to preserve the genetic diversity of Vanilla crop wild relatives (CWRs) while simultaneously mitigating the impact of extreme climates on vanilla cultivation.

Results

Clonal assignment and estimates

Among the 75 V. odorata individuals collected in the Osa Peninsula, we identified 26 unique genets across all populations. Vanilla odorata has an average clonality of 63%. There is great variation in the level of clonal reproduction among populations, with Miramar and VT having the largest levels of clonality (75%). According to Simpson’s Diversity Index (SDI), the probability of finding two individuals with different genotypes is 25% in Miramar, while in Matapalo it rises to 80%. The average evenness is 0.64, indicating moderate genotypic diversity, as the distribution of genotypes within populations is not completely equitable, but neither is it extremely uneven. Matapalo exhibited the highest evenness (0.93), suggesting a homogeneous frequency distribution of all genotypes. In contrast, VT had the lowest evenness (0.43), indicating that a single genotype is predominant throughout the population (Table 1, Supplementary Fig. S1).

Table 1 Clonal diversity estimates for V. odorata populations in Southern Costa Rica.
Full size table

Among the 146 V. pompona individuals collected in the Pacific Region of Costa Rica, we identified a total of 93 unique genets. Vanilla pompona has significantly less clonality, with an average clonality of 35%. Similar to V. odorata, there is significant variation in clonality across populations, with VT exhibiting the highest levels at 75%. According to the Simpson’s Diversity Index, the probability of finding two different genets in VT is 66%, while in Rio Jaris the probability of finding two individuals with different genets is 100%. Average evenness is 0.74, suggesting a balanced and diverse meta-population, with multiple genotypes present in relatively similar proportions (Table 2).

Table 2 Clonal diversity estimates for Vanilla pompona populations across the Pacific region of Costa Rica.
Full size table

Furthermore, our results show that the average maximum clonal patch size in V. odorata was 36.2 (± 13.9) meters (Supplementary Fig. S1). For V. pompona, the average maximum spread of clones was 28.4 (± 12.5) meters (Supplementary Fig. S2). An intermixed pattern of genets was observed in both species, with different clones and genets coexisting in the same area. However, in V. odorata populations such as Miramar, VT, and PST, a few genets dominate, creating large clonal patches that reach up to 10 m in Miramar and at least 74 m in VT (Supplementary Fig. S1). A similar pattern was observed in V. pompona, although with greater clonal diversity (SDI) within populations. Only a few populations had large clones, such as VT, which had two genets that extended up to 32 m, and Los Chocuacos, where one genet extended up to 31 m. In the case of El Hacha, the same genet was identified in two individuals separated by up to 112 m.

Genetic diversity

All markers in both species deviated from Hardy-Weinberg equilibrium (HWE), as expected for clonal species. Across the 10 loci analyzed in V. odorata, the number of alleles ranged from two (mVplCIR047) to six (mVhuCIR07). In V. pompona, the 11 loci analyzed exhibited a broader range, with three to 15 alleles, with mVhuCIR07 being the most polymorphic. Our estimates of genetic diversity were similar between populations within each species, with few differences between analyses including and excluding clones, except possibly in FIS. Genetic diversity estimates were slightly influenced by sample size. For example, in V. odorata in Miramar, we found the lowest diversity metrics because it is the population with the smallest sample size. The two populations with the fewest genets (VT and Garabito) also had the lowest genetic diversity estimates in V. pompona. Vanilla odorata shows lower diversity metrics than V. pompona, with fewer alleles (A = 1.70 vs. 2.76), lower allelic richness (Ar = 1.57 vs. 2.30), and a smaller effective number of alleles (Ae = 1.57 vs. 2.09). Although V. pompona has higher expected heterozygosity (He = 0.40, 95% CI = 0.20–0.58) compared to V. odorata (He = 0.27, 95% CI = 0.07–0.45), confidence intervals overlap. There is no clear geographic pattern in the distribution of genetic diversity. Inbreeding in V. pompona was low and not significant (FIS = -0.03, SE = 0.19), while a significant heterozygote excess was found in V. odorata (FIS = -0.67, SE = 0.17) (Tables 3 and 4).

Table 3 Genetic diversity estimates for six populations of V. odorata in Costa Rica, based on a dataset including clones (IC) and excluding clones (EC). N sample size, A average number of alleles per locus, Ar allelic richness, Ae effective number of alleles, Ho observed heterozygosity, He expected heterozygosity, FIS inbreeding coefficient. Mean values ± (SE). *Indicates significant FIS (P < 0.01) based on 1000 bootstrap replicates.
Full size table
Table 4 Genetic diversity estimates for 10 populations of V. pompona in Costa Rica, based on a dataset including clones (IC) and excluding clones (EC). N sample size, A average number of alleles per locus, Ar allelic richness, Ae effective number of alleles, Ho observed heterozygosity, He expected heterozygosity, FIS inbreeding coefficient. Mean values ± (SE). *Indicates significant FIS (P < 0.01) based on 1000 bootstrap replicates.
Full size table

Genetic structure and isolation by distance

We found significant genetic structure at the population level in both species: GST = 0.514 (95% CI = 0.500–0.529) for V. odorata; and GST = 0.091 (95% CI = 0.075–0.108) for V. pompona. Structure was significantly higher in V. odorata. The DAPC analysis of V. odorata shows a clear separation of populations in the inner part of the peninsula and those populations in Piro, forming a separate cluster (Fig. 1A). We found significant IBD for V. odorata (rIC = 0.704, p < 0.05; rEC = 0.737, p < 0.05; Supplementary Fig. S3).

Fig. 1
figure 1

Discriminant analysis of principal components (DAPC) (A) clustering of six populations of V. odorata, including clones (IC). (B) clustering of ten populations in V. pompona, including clones. Colors denote the geographical allocation of V. pompona populations (clusters) to designated regions: Green = South Pacific - Peninsula; Yellow = South Pacific - Térraba; Pink = Central Pacific - Carara; Blue = North Pacific; Purple = Central Pacific - Puriscal.

Full size image

In V. pompona, the DAPC analysis revealed a very low level of genetic structure, resulting in most populations clustering together. Only two populations, STH and VT, stand out as genetically distinct. Populations from the South Pacific–Térraba (Clavera and Los Chocuacos), Central Pacific–Carara (Pavona and Garabito), North Pacific (El Hacha), and PST all form a central cluster, reflecting similar allele frequencies (Fig. 1B, Supplementary Fig. S6). Additionally, San Rafael and Rio Jaris form a separate but closely related cluster to the main group (Fig. 1B, Supplementary Fig. S6). No significant IBD was found among V. pompona populations (rIC = 0.138, p = 0.208; rEC = 0.056, p = 0.275; Supplementary Fig. S4) or at the regional level (rIC = 0.351, p = 0.187; rEC = 0.533, p = 0.123; Supplementary Fig. S5).

Fine scale spatial genetic structure

For both species, pairwise relatedness decreased with increasing pairwise distances between individuals (Figs. 2 and 3). The point at which the correlogram crossed the X-axis allowed us to calculate the average patch size26,27. In V. odorata, the average patch size was estimated at around 4120 m for both datasets (Fig. 2). For V. pompona, average patch size was approximately 92 m when clones were included and 283 m when clones were excluded (Fig. 3). Autocorrelation between V. odorata individuals separated by less than 5 m, including ramets (r = 0.498; Fig. 2A), was significantly different from zero and twice as large as for our data set with only genets (r = 0.249; Fig. 2B). In V. pompona, autocorrelation between individuals separated by less than 3 m, including ramets (r = 0.250; Fig. 3A), was significant and slightly higher than for individuals separated by less than 4 m in the genets subset (r = 0.190; Fig. 3B). However, spatial autocorrelation remains significant and high in V. odorata up to 1560 m, and up to 60 m and 243 m in V. pompona when clones are included (IC) and excluded (EC), respectively (Figs. 2 and 3). Loiselle’s coancestry coefficient was consistently higher for the ramet dataset than for the genet dataset, indicating a strong influence of clones in shaping spatial genetic structure in the closest distance classes. When FSGS analyses were performed by population for each species, similar patterns were observed (Supplementary Figs. S7 and S8). However, individual autocorrelation was lower because larger distance classes were needed due to the limited number of individuals within populations.

Fig. 2
figure 2

Correlograms showing the spatial autocorrelation of the coancestry Loiselle coefficient (Loiselle et al., 1995) as a function of distance for V. odorata populations: (A) dataset including clones, (B) dataset excluding clones. Dashed lines represent the lower and upper 95% confidence limits for the null hypothesis of no spatial genetic structure.

Full size image
Fig. 3
figure 3

Correlograms showing the spatial autocorrelation of the coancestry Loiselle coefficient (Loiselle et al., 1995) as a function of distance for V. pompona populations: (A) dataset including clones, (B) dataset excluding clones. Dashed lines represent the lower and upper 95% confidence limits for the null hypothesis of no spatial genetic structure.

Full size image

Discussion

In our studies, the clonal nature of V. odorata and V. pompona has proven to be a critical factor in understanding the amount and distribution of genetic diversity. In comparing V. odorata and V. pompona, our study highlights sharp differences in clonality, geographic distribution, genetic diversity, and structure. In Costa Rica, while V. odorata is more clonal and has a restricted distribution, primarily confined to the Osa Peninsula, V. pompona has a broader distribution along Costa Rica’s Pacific coast. As expected, V. pompona showed higher levels of genetic diversity compared to V. odorata, likely due to its wider distribution, reduced clonality, and increased opportunities for sexual reproduction and gene flow. This is confirmed by V. pompona’s lack of inbreeding and low levels of spatial genetic structure in contrast with V. odorata. The restricted and clonal nature of V. odorata populations is consistent with our predictions, as it results in heterozygote excess due to its limited genetic diversity and predominantly vegetative propagation.

In our results, repeated genotypes were considered products of vegetative reproduction, particularly guerrilla-type growth, where ramets are widely spaced and rapidly spread into new areas28. Vegetative or clonal reproduction in orchids often correlates with reduced genetic diversity, as observed in Vanilla planifolia6,7. Similarly, the orchid Pelatantheria scolopendrifolia (Makino) Aver. had moderate to low levels of genetic diversity, attributed primarily to small population sizes, demographic events, clonal reproduction, and limited gene flow29. Clonal architecture, which influences the spatial distribution of ramets, can significantly impact geitonogamy30. In populations of V. odorata, such as Miramar, Tigre, VT, and PST, a clumped distribution of ramets was observed. In V. odorata, this clustering will likely lead to a higher degree of geitonogamy30 and may contribute to the significant genetic structure observed among populations. Conversely, V. pompona individuals are typically more intermingled (except in VT), which may lead to higher outcrossing rates30. A mixed distribution of genets can improve outcrossing by increasing the availability of diverse pollen from multiple genets30. This pattern is commonly observed in populations of V. pompona (Supplementary Fig. S2).

The near-zero inbreeding coefficient observed in V. pompona (FIS = -0.03) is in direct contrast with high inbreeding values reported for V. mexicana (FIS = 0.74), where autogamy is prevalent16. Thus, the higher levels of genetic diversity observed in V. pompona are most likely due to relatively higher levels of sexual reproduction compared to clonal growth. In contrast, V. odorata’s heterozygote excess (FIS = -0.67) may be attributed to its high clonal reproduction rates that maintain large fractions of heterozygous ramets within populations. This pattern is also consistent with bottleneck or founder effects, where guerrilla propagation of clones may lead to the widespread establishment of heterozygous genotypes31. Clonal reproduction may maintain heterozygosity or even increase it by mutation over generations32,33.

The genetic diversity observed in Vanilla odorata and V. pompona aligns with patterns observed in other orchids, with diversity levels influenced by the degree of clonality. In V. odorata, smaller and highly clonal populations, such as Miramar, Tigre, and Matapalo, showed notably lower heterozygosity values compared to larger, less clonal populations like PPT and PST. Similarly, in V. pompona, the more clonal populations VT, PST, and Garabito had reduced genetic diversity, whereas larger populations like STH, Clavera, and La Pavona exhibited higher genetic diversity. Previous studies have shown that in clonal orchids, such as Cypripedium calceolus L., populations that rely more heavily on clonal reproduction tend to have lower genetic diversity due to limited genetic recombination compared to populations that favor sexual reproduction34. When compared with other Vanilla species, V. odorata exhibits low to moderate genetic diversity, which aligns with findings for V. planifolia, another highly clonal species with a deceptive pollination system21. Vanilla pompona has higher levels of genetic diversity, similar to African Vanilla species (V. madagascariensis, V. decaryana, V. humblotii, V. perrieri, and V. bosseri), which combine both clonal and sexual reproductive strategies but tend to have lower clonal reproduction15,35. Like V. pompona populations in Costa Rica, V. madagascariensis, V. decaryana, and V. perrieri also tend to have larger populations and broader distributions across Madagascar, contributing to their higher genetic diversity. Together, these results underscore how reproductive strategy and population size can influence genetic diversity, with clonality often associated with reduced diversity due to limited genetic recombination and smaller effective population sizes36,37.

Gene flow within Orchidaceae has been shown to occur over distances of up to 250 km38, but greater population differentiation can arise when intermediate populations are scarce, limiting gene flow between distant populations38. The genetic structure in Vanilla odorata and V. pompona likely reflects the combined influence of limited gene flow through pollen and seed dispersal, shaped by distinct pollination and dispersal mechanisms in each species. In V. odorata, a higher population structure is observed, likely due to pollination through food deception by both male and female Euglossa bees21. In deceptive systems, pollinators receive no reward (such as food or fragrance), so they quickly learn to avoid these flowers, resulting in less frequent visits and, consequently, fewer opportunities for effective pollination39. Limited pollen dispersal in V. odorata, combined with short-range seed dispersal by fragrance and pulp collection by Euglossa bees and Trigona fulviventris (Meliponini)40, likely contributes to the Isolation by Distance (IBD) pattern and strong genetic clustering, particularly in geographically isolated populations like Tigre, Miramar, and Matapalo.

Conversely, V. pompona, which is distributed along the Pacific coast, exhibits lower genetic structure despite its broad range. This pattern can be explained by the stepping-stone theory, where populations connected by intermediate groups maintain genetic connectivity across larger distances, even if direct gene flow between distant populations is limited41,42. In V. pompona, Eulaema cingulata was identified as the main pollinator, engaging in floral scent collection and nectar searching43. This behavior facilitates pollen removal and attracts bees from long distances, potentially increasing gene flow43. In addition, V. pompona benefits from effective seed dispersers like the spiny rat (Proechimys semispinosus) and agouti (Dasyprocta punctata), which have extensive home ranges (up to 1,258 m during the rainy season) and play a significant role in maintaining genetic connectivity across Central American forests44. These dispersers help spread seeds over greater distances, supporting high genetic connectivity in V. pompona despite geographical separation. Overall, these findings highlight how biological interactions, as well as clonality, population size and gene flow, influence genetic structure in V. odorata and V. pompona.

Interspecific gene flow via hybridization may also contribute to genetic diversity and structure estimates in our samples, as V. pompona and V. odorata are sympatric. Although natural hybrids between V. pompona and V. odorata may exist, their occurrence under natural conditions is highly unlikely due to the specificity in plant-pollinator relationships and the specialized pollination strategies in Vanilla. As previously stated, the pollinators of V. pompona are male Eulaema, a large bee that is attracted to the flowers by floral fragrances, which are the main reward43. Conversely, the deceptive V. odorata is likely pollinated by smaller Euglossa or Meliponini bees, analogous to the pollination of the florally similar V. planifolia21. Furthermore, in natural conditions, interspecific pollen transfer is highly unlikely due to the lack of flowering overlap between both species43,45. Considering the morphological differences of pollinators and the lack of synchrony in flowering phenology, natural hybridization between V. odorata and V. pompona is unlikely.

Fine-scale genetic structure (FSGS) in plant populations is defined as the non-random distribution of genotypes at a local scale, which can be caused by limited seed dispersal, clonality, or environmental factors46. In V. odorata, the strong genetic structuring observed when clones were included, suggests that clonality plays a dominant role in shaping local genetic patterns, as reported in previous studies14,36. However, after removing clones, the genetic structure revealed a pattern consistent with full-sibling relationships47, indicating that seed dispersal in sexually reproduced progeny is highly localized. In V. pompona, a similar but less clear pattern was observed. The moderate relatedness observed when clones were included, indicates a mixture of clonal individuals, sexually produced progenies, and unrelated individuals. Once clones were excluded, relatedness decreased significantly but remained significant up to 240 m, indicating large clusters of related individuals composed of full- and half-siblings within V. pompona populations. A similar pattern was observed in V. humblotii Rchb.f., where FSGS was greater for the ramets (datasets with clones) than for the genets alone, suggesting that, like in V. odorata and V. pompona, the repeated multilocus genotypes (clones) play a significant role in defining a strong spatial genetic structure, particularly in the first distance classes14.

Despite V. pompona’s potential for long-distance gene dispersal, many seeds may remain in close proximity, leading to local FSGS (see Fig. 3). Studies regarding the seed dispersal of Neotropical Vanilla suggest, that although potentially being able to have long-distance dispersal, most seeds are likely being dispersed relatively short distances from the plant40,48. Specifically, Vanilla pompona fruits are indehiscent, naturally dropping below the plant when mature. The fruit is then consumed fully or in part by mammals, especially rodents, which can disperse the seeds up to 18 h after consumption, usually within its home range40. In the case of Vanilla odorata fruits are dehiscent, naturally exposing the seeds that are displaced by fragrance collecting male orchid bees, causing seeds to drop directly below the fruit40. This localized seed dispersal, coupled with the behavior of seed vectors, reinforces the development of fine-scale genetic structure in both V. pompona and V. odorata.

Our findings highlight that both population size and clonality significantly shape genetic diversity in V. odorata and V. pompona. Additionally, clonality increases genetic structuring, particularly in small, isolated populations. From a conservation perspective, our study underscores the importance of maintaining larger population sizes and reducing isolation to support genetic diversity, especially in clonal species like V. odorata, where genetic diversity is already limited. Preserving diverse and interconnected habitats for V. pompona could enhance gene flow, helping to maintain population sizes and resilience. For V. odorata, focusing on reducing genetic bottlenecks by protecting diverse habitats and promoting occasional outcrossing could mitigate genetic drift and support long-term viability. These strategies are essential for conserving these unique genetic resources and the adaptive potential of both species in Costa Rica’s tropical forests. To achieve this goal, integrated conservation strategies recommended by Watteyn et al. (2020) include protecting forest areas where wild vanilla species naturally occur, promoting reforestation of biological corridors, cultivating vanilla alongside native tree species with economic benefits, and implementing payments for ecosystem services to incentivize farmers and landowners to conserve native vanilla populations while adopting sustainable agricultural practices11.

Moreover, our results suggest that V. pompona may be an important resource to enhance the genetic diversity of cultivated vanilla through breeding strategies that include hybridization with V. planifolia. This approach could promote sexual reproduction while maintaining high genetic diversity. The broad distribution and reduced genetic structure of V. pompona make it a valuable resource from which individuals can be selected from various sites. Preliminary studies indicate that wild V. pompona populations possess desirable traits for commercial plantations, such as increased resistance to pathogens like Fusarium oxysporum f. sp. vanillae and increased drought tolerance, compared to cultivated V. planifolia49,50. Menchaca et al. (2011) confirmed the feasibility of generating hybrids between V. planifolia and V. pompona, which exhibited high germination rates and remarkable hybrid vigor51. On the other hand, it is crucial to note that when collecting individuals of V. odorata from patches of 30–40 m, there is a high likelihood that they belong to the same clone. This highlights the importance of considering spatial genetic structure when collecting and using wild populations for breeding programs. These findings emphasize the importance of conserving multiple populations to preserve the overall genetic diversity of the species.

While this study employed microsatellites and included a robust sampling effort, certain limitations should be acknowledged. The limited number of populations and markers, particularly for Vanilla odorata, a rare species confined to just a few localities in Costa Rica, may influence the scope of our findings. For future research, we recommend expanding marker types and genomic coverage, such as using SNPs, as well as incorporating nuclear and organellar markers like plastomes to provide complementary insights into the species’ evolutionary history. Additionally, including more populations across a broader geographic range and integrating landscape genetics would help to further explore how geographic and environmental factors shape genetic variation.

Methods

Study species

Vanilla is a pantropical genus consisting of around 140 species of hemiepiphytic and epiphytic vines within the Orchidaceae family2. Vanilla odorata and V. pompona are distributed from Mexico to Peru and Brazil45, with both species present in Costa Rica’s Osa Conservation Area (ACOSA)11. Vanilla pompona also occurs along the Pacific slope of Costa Rica. Their flowers consist of three sepals, two lateral petals, and a large trumpet-shaped petal known as the labellum, which is partially fused to the column. This leads to a single ventral stigma, separated by the rostellum from the anther, which holds two pollen masses45. Vanilla pompona blooms in Costa Rica from January to February43, with flowers measuring 6 to 8 cm in length and 7.5 to 10 cm in diameter. The successive inflorescences have one to two open flowers at a time, which are ephemeral, lasting only one day and wilting after noon; they are pale yellow with a striking yellow-orange labellum and a strong mint-like fragrance45. In contrast, Vanilla odorata flowers bloom in April and May45, and are 8 cm long and 7 cm wide45. They feature successive inflorescences with one to two open flowers at a time, ephemeral from 7:00 a.m. to 4:00 p.m., with pale greenish-white tepals and a greenish-white labellum, and have a mild, fresh fragrance45.

Sample collection

The localities cited by Karremans et al. (2020) for V. odorata and V. pompona were visited and sampled in the current study, with the materials collected being highly representative of the known populations of both species in the country. Vanilla plants used for this study were identified by A.P. Karremans, following the circumscription of Karremans et al. (2020). Representative vouchers for each study site for both species are kept as living plants or liquid specimens at Centro de Investigación Jardín Botánico Lankester (JBL), in Cartago, Costa Rica. A complete list of voucher numbers is provided in Supplementary Table S1. We selected 10 V. pompona populations along the Pacific slope of Costa Rica (Fig. 4). For V. odorata, we collected samples from six populations located in the Osa Peninsula (Fig. 4). We sampled between 8 and 20 individuals from each of the 10 V. pompona populations, totaling 146 individuals; these individuals were grouped into five regions (See Fig. 1). For V. odorata, we sampled from 5 to 24 individuals from each of the six populations. Each individual was georeferenced, and we collected one to two leaves per individual, which were stored in silica gel and processed within three days at the Laboratorio de Ecología Molecular of the School of Biology at Universidad de Costa Rica. All procedures complied with Costa Rican laws, and research permits were approved by the Research Committee of the School of Biology at the University of Costa Rica under project approval code C0049 from the Vice-Rectorate of Research.

Fig. 4
figure 4

A map of Costa Rica showing the locations of V. odorata (triangles) and V. pompona (circles) populations. The colors grouping the circles in V. pompona correspond to the five regions in which the populations are categorized: Blue = North Pacific; Pink = Central Pacific - Carara; Purple = Central Pacific - Puriscal, Yellow = South Pacific - Térraba Green = South Pacific - Peninsula. The numbers in brackets indicate the number of individuals collected at each site. In the zoomed-in view of the Piro Biological Station, the names in light orange represent V. odorata populations, while the names in light green represent V. pompona populations. Abbreviations are as follows: VT = Vanilla Trail, PST = Piro Station Trail, PPT = Pica Pica Trail, STH = Stairway to Heaven. The map was created using QGIS software (version 3.30.3-‘s-Hertogenbosch, available at https://qgis.org/), with minor modifications made in Adobe Photoshop® CC (Adobe Systems Inc., California, U.S.A.). QGIS is licensed under the GNU General Public License (https://www.gnu.org/licenses).

Full size image

DNA extraction and SSR amplification

We extracted total DNA from leaf tissue using the modified CTAB protocol from Doyle and Doyle (1990)52. Dilutions were prepared to achieve a final concentration of 20 ng/µL for PCR reactions. Microsatellites (SSRs) previously proposed in the literature for other Vanilla species were used in this study6,17. A total of 28 SSR markers were tested for transferability, of which 16 successfully amplified in V. pompona and 15 in V. odorata. Among these, 12 markers were polymorphic for each species. However, one marker was excluded from the analysis in V. pompona and two in V. odorata due to missing data. Therefore, eleven microsatellite loci were used for V. pompona: mVhuCIR07, mVhuCIR08, mVhuCIR10, mVhuCIR11, mVroCIR01, mVroCIR05; mVplCIR015, mVplCIR016, mVplCIR019, mVplCIR026, mVplCIR031; while 10 microsatellite loci were used for V. odorata: mVhuCIR07, mVhuCIR08, mVroCIR01, mVroCIR03, mVroCIR05; mVplCIR05, mVplCIR025, mVplCIR026, mVplCIR031, mVplCIR047. The forward primer of each primer pair was fluorescently labeled at the 5’ end. PCR thermal profiles and primer mixes are described in Supplementary Table S2 of the supplementary material. We used a Veriti™ thermal cycler (Applied Biosystems, Foster City, CA, USA) to perform the PCR reaction. Amplification products were visualized in an ABI-3500 Sanger sequencer at Escuela de Biología at Universidad de Costa Rica, using Hi-Di™ Formamide and GeneScan™ 500 LIZ™ dye Size Standard (Applied Biosystems). Genotypes were scored manually using GeneMarker Software version 2.6.4 (SoftGenetics).

Clonal assignment

We used GenoDive v. 3.0653 to determine the number of genets and associated ramets in each population. For V. pompona, the dataset included 146 individuals and 10 of 11 loci (mVplCIR016 was excluded due to missing data). For V. odorata, the dataset comprised 75 individuals and all 10 loci. A Stepwise Mutation Model was applied in both analyses, excluding missing data from the counts. For V. pompona, the minimum genetic distance threshold for distinguishing different clonal lineages was established at 1, whereas for V. odorata, the threshold was set at 054 and the “Make clones specific to every population” option was selected for the analyses in both species. These thresholds were estimated from a frequency distribution of genetic distances, assuming that the first peak in the histogram was likely due to somatic mutations or genotyping errors55. The threshold was chosen between the first peak and the next increase in genetic distance frequencies. To assess clonal structure, we tested the probability of observing the detected clonal diversity by selecting the number of genotypes as the statistic for both species. We conducted 9,999 permutations, with randomization set to “alleles over individuals within populations” for V. pompona and “randomize alleles over individuals across all populations” for V. odorata.

We estimated the number of genets and calculated the clonality percentage by dividing the number of genets by the total number of individuals in each population. We also calculated Simpson’s Diversity Index (SDI), which measures genetic diversity by quantifying the probability that two randomly selected individuals belong to different genotypes. Additionally, we assessed evenness (E), which indicates the uniformity of genotype distribution among sub-populations. An evenness value of 1 indicates that all genotypes are equally represented, implying balanced diversity, whereas lower values indicate an uneven distribution, with some genotypes dominating. All these estimates were calculated for each population using GenoDive v. 3.0653. The analysis involved a bootstrap test for differences in clonal diversity with 9999 permutations and a jackknife procedure with increasing sample sizes, using 1000 permutations.

Genetic diversity and structure

Genetic variation and its distribution within populations of V. pompona and V. odorata were described by estimating the average number of alleles (A), allelic richness to compensate for differences in sampling size per population (Ar), effective number of alleles (Ae), average observed (HO) and expected (He) heterozygosity, and the inbreeding coefficient (FIS). We tested each locus for significant deviations from Hardy-Weinberg equilibrium (HWE) using the exact test implemented in the Genepop software56. We estimated the significance of inbreeding coefficients using 1000 bootstrap samples via the mmod package57. To evaluate differences in allele frequencies among populations, we calculated the Nei’s GST index58. Isolation-by-distance (IBD) was measured by comparing Euclidean distances between populations using Nei’s GST, and the significance of IBD was determined using a Mantel test with 9,999 permutations. All genetic diversity and structure estimates were calculated using the adegenet59 and poppr54,60 packages in R version 4.2.361. All genetic diversity estimates were calculated in two separate analyses, one including clones (IC) and a second analysis excluding clones (EC) within each population (see sub-section Clonal assignment).

We visualized the dissimilarity among individuals and populations using a Discriminant Analysis of Principal Components (DAPC) as implemented in the adegenet package59. Cross-validation was conducted by partitioning the data into a training set (75%) and a validation set (25%), repeating this process 1,000 times at each step of Principal Component (PC) retention. To determine the optimal number of PC axes, we used the xvalDapc() function from the adegenet package with 30 repetitions. We then refined our search to the five PC axes closest to the previously selected PC, increasing the number of replicates to 999. We retained 25 components and three discriminant functions for V. pompona DAPC, while in V. odorata we retained seven components and three discriminant functions.

Fine scale spatial genetic structure (FSGS)

Autocorrelation analyses based on multilocus pairwise kinship coefficients47 were calculated using SpaGeDi 1.562. We selected distance classes with at least a minimum of 30 pairwise comparisons per distance class63. Fine-scale spatial genetic structure (FSGS) analyses were conducted globally for each species, combining all sampled populations into a single analysis per species. These analyses were conducted both with clones included (IC) and excluded (EC), which resulted in different distance intervals for each set of analyses. In V. odorata, the IC dataset had 19 distance classes, while the EC dataset had seven. For V. pompona, 33 distance classes were defined in the IC dataset, compared to 15 in the EC dataset. We also ran separate FSGS analyses on each population to see if patterns were consistent across populations.

Using the IC datasets, we calculated the coancestry coefficient ρij to measure FSGS within populations that had at least 15 individuals47. For V. pompona, individual analyses were performed on six populations (VT, STH, La Pavona, Los Chocuacos, Clavera, and San Rafael), while for V. odorata, only three populations (PST, VT, and PPT) were analyzed. Permutation tests were conducted to evaluate the statistical significance of the observed patterns, with 10,000 permutations assigned to each test.