We inventoried ‘Medicinal Orchids of Asia and Orchids as Aphrodisiac, Medicine, or Food’ (Teoh, 2019; Teoh, 2016), and selected orchids of which the tubers, or other parts such as pseudobulbs, were described as edible and having uses or properties that were similar to salep, for example, to soothe coughs. This was supplemented by references focused on smaller regions or specific countries (Challe and Price, 2009; Fuchs, 1542; Ghorbani et al., 2014; Gras et al., 2017; Hamisy, 2010; Kasparek and Grimm, 1999; Kasulo et al., 2009; Kreziou et al., 2016; Lee et al., 2008; Low, 1987; Moerman, 2021; Tekinşen and Güner, 2010; Uprety et al., 2016; Veldman et al., 2017, 2018). Reference source, name in reference, and use were recorded alongside currently accepted name and, when documented, chromosome number (Cozzolino et al., 2006; Leitch et al., 2019; POWO, 2024; Rice et al., 2015). This resulted in a list of 203 taxa (See Supplementary Data Appendix I).

For each orchid selected, we checked records of cross-compatibility by searching parentage records, as pollen parent and seed parent, against all other orchids within our selection in The International Orchid Register (Royal Horticultural Society, 2024). This was used to generate a matrix of known hybrids (Supplementary Data Appendix II).

Multispecies sequence alignments were made with DNA sequences of nrITS and matK markers generated by Jin et al. (2017) downloaded from NCBI GenBank, and aligned using MAFFT (Katoh and Standley, 2013). We followed the methods described in Viruel et al. (2021) to reconstruct an ultrametric phylogenic tree and calculate pairwise phylogenetic distances. Phylogenetic trees were reconstructed using IQ-TREE applying the GTR + G model of nucleotide evolution and 1,000 replicates of bootstrapping searches (Nguyen et al., 2015). An ultrametric tree was produced using treePL by assigning the value of ‘1.0’ to the root of the tree (Smith and O’Meara, 2012). The ultrametric tree was used to calculate pairwise phylogenetic distances using the patristic method as implemented in the function distRoot from the R package adephylo (Jombart et al., 2010; Pavoine et al., 2008). Species from Cranichideae and Diurideae were used as outgroups. Both Cranichideae and Diurideae are predominantly terrestrial tribes of orchids, of which no species are used as salep (Salazar et al., 2003; Zhang et al., 2023). Phylogenetic trees were visualized with iTOL v6 (Letunic and Bork, 2024). For the heat map analyses, a total of 17 intergeneric hybrids registered with the Royal Horticultural Society (Table 1), were scored for whether both, or one of the two parental species were present in the alignments. If no parental species was present in the dataset, a score of 0 was assigned, if one of the parents was sequenced a score of 1 was given, and if both parental species were sequenced a 2 was assigned. The consensus tree was visualized using the ape package (Paradis and Schleip, 2019) and the ggtree package (Yu et al., 2017), with the heatmap generated by the gheatmap function in R (R Core Team, 2023) within RStudio (RStudio Team, 2022). This approach included an integrated heatmap representing the relationships between nothogenera and their parental species as tree tip labels. The same method was employed to visualize a heatmap of patristic distances derived from the phylogenetic analyses.

Hybrid crossing attempts were conducted in the Netherlands in collaboration with horticultural hobbyist growers from the Vereniging Orchideeën Vermeerdering (VOV) Working Group on salep. Pollinia were removed from the paternal parent using tweezers and placed on the stigma of the maternal parent, whose pollinia had also been removed and discarded. Pollinia packages from 1 to 4 fresh flowers from 6 father plants of a total of 4 species were hand-collected and transferred to the stigma of flowers on 6 mother plants in 2020 and 2023 (Table 2). This resulted in the formation of 6 sets of fruits in 2020, and 2 sets of fruits in 2023 (see Table 2).

Choice of parentage of the hybrid was based on having both parent plants verifiably cultivated and legally available for use, and having desirable tuber characteristics such as fragrant tubers (Dactylorhiza maculata subsp fuchsii).

In 2020, for two crosses, we used Dactylorhiza x foliorella, a hybrid that is readily available in cultivation, as the maternal and paternal parent, respectively. This hybrid cannot occur spontaneously in the wild due to geographical separation of its parental species: Dactylorhiza foliosa is endemic to Madeira, and D. purpurella is native to Denmark, Great Britain, the Netherlands, and Norway (POWO, 2024). We crossed D. × foliorella with Dactylorhiza fuchsii as the maternal and paternal parent, respectively, one of the orchids used in salep that has particularly fragrant tubers and the characteristic palmate tubers of Dactylorhiza.

Serapias is a genus of orchids with known capacity to produce more than one new tuber per year, individual plants have 2 to 5 tubers (Delforge, 2005). Additional crosses were attempted between Serapias cordigera and Dactylorhiza fuchsii. Tuber production by Anacamptis morio has been documented in horticulture (Cribb and Bailes, 1989). In 2023, we crossed A. morio with two species of Serapias: S. strictiflora and S. lingua. A. morio tolerates a range of growth conditions from dry to damp and acidic to alkaline soils, and is found from North Africa to the boreal (Plant Atlas 2020, 2024; POWO, 2024). S. strictiflora and S. lingua form more than one new tuber, and have not been reported as being traded for salep although other species of this genus are traded for salep (Masters et al., 2022; Veltman et al., 2023).

Pod ripening was assessed visually. Nearly ripe (still closed) pods were sent to Phytesia SA, an orchid growing company, in 2020 and members of the Vereniging Orchideeen Vermeerdering in 2023. Seeds were sown on nutrient medium. DNA was extracted from fresh leaves of both parental plants and hybrids following the CTAB protocol of Doyle and Doyle (1990) with overnight incubation in CTAB buffer for 1–3 days. PCR on a Westburg T3 Biometra thermocycler was used to amplify plastid trnL-trnF with a and b, c and d, e and f primers (Taberlet et al., 1991) and nrITS1 and nrITS2 DNA barcoding markers using 17SE and 26SE primers (Sun et al., 1994). NucleoSpin Gel and PCR clean-up kits (Sigma-Aldrich) were used to purify all resulting amplicons. A total of 5 microliter of PCR product was loaded on an 1.5% agarose gel, visualized by staining with Ethidium bromide, and sized by visual comparison with a standard size ladder. PCR products of the correct size were sent to be Sanger sequenced on an ABI 3730 XL (ThermoFisher) at Radboudumc Technology Center Genomics in Nijmegen. (See Supplementary Data Appendix III gel electrophoresis images of hybrid amplicons). The DNA sequences generated for hybrid validation were not included in the phylogenetic framework.

Pairwise phylogenetic distances were calculated (see Supplementary Data Appendix IV) and a consensus phylogenetic tree of orchids with known or potential use as edible was generated (see Supplementary Data Appendix V). Inter-generic hybrids registered on the International Orchid Register were combined with the phylogenetic tree to generate a heatmap of crosses between species, and crosses between genera for registered inter-generic crosses. The combined phylogenetic tree and heatmap generated is depicted in Figure 1.

The combined phylogenetic tree and heat map show that phylogenetic distance can be used as a proxy for predicting hybrid compatibility. Aggregation is visible and already registered, or potential hybrids are not spread all over the phylogeny but clearly confined to specific clades of closely related species with low phylogenetic distances. Intergeneric compatibility seems to be restricted to the genera Anacamptis, Cynorkis, Dactylorhiza, Gymnadenia, Habenaria, Himantoglossum, Neotinea, Orchis, Ophrys, Pecteilis, Platanthera, Pseudorchis, and Serapias. There are several intergeneric hybrids with underexploited potential. Amantoglossum (Anacamptis x Himantoglossum) and Orchimantoglossum (Himantoglossum x Orchis) may be particularly desirable for cultivation because Himantoglossum is a genus known for large tubers offering the possibility of high yield. There was no Serapirhiza (Serapias x Dactylorhiza) hybrid registered for the species within our phylogenetic tree, however the two genera are compatible. So, attempting to hybridize species of Dactylorhiza and Serapias known to be traded as salep is likely to succeed.

Validation via DNA barcoding rejected successful hybridisation for 5 out of the 6 crosses attempted but confirmed that a hybrid was successfully created between Dactylorhiza × foliorella and D. maculata subsp. fuchsia, i.e., the seeds harvested from the pods were not the result of self-pollination. This newly created hybrid was registered with the International Orchid Register (Royal Horticultural Society, 2024) under P.34144. It is also documented via botanical illustrations and photographs (see Figure 2). Botanical illustration of hybrid ‘Sarah’ and Figure 3 photograph of flowers of hybrid ‘Sarah’. A description of this newly created hybrid and the three species of its parentage, highlighting morphological differences between the hybrid and its parents, was created (Supplementary Data Appendix VI).

Known and likely intergeneric hybrids clustered within close proximity on the phylogenetic tree. This confirms that intergeneric hybrids can occur but are more likely between closely related genera. Our integrated heatmap and phylogenetic tree show the territory of known hybrids and highlight species with edible tubers that are likely to successfully hybridize. This predictive power is useful for focusing hybridisation efforts where they are most likely to succeed. This is particularly important when the time scale for hybridisation is both constrained by seasonal flower production, and long due to the length of time from flowering to seed capsule formation, and then germination, followed by growth to maturity.

Our successful hybridisation of D. x foliorella crossed with Dactylorhiza maculata subsp. fuchsii, combined a species known to have particularly fragrant tubers with a hybrid containing the germplasm of an endemic species not currently in salep trade. This created a hybrid with distinct morphology, that cannot occur as a wild plant and is also predisposed to meeting requirements for use as salep.

To prevent the risk that our hybrid, or others, out-compete or back cross with wild species, additional steps need to be taken, such as developing infertile plants with peloric flowers, in which the inner two floral whorls with stamen and stigma are replaced by tepals, so-called non-perianth transitions (Bateman and Rudall, 2006). Such sterile ‘double flowers’ have been developed for many horticultural varieties, for instance, rose, peony, and carnation (Meyerowitz et al., 1989), and are sporadically found in wild orchids (Figure 4). A potential additional benefit from sterile flowers is that resources may be directed to tuber growth instead of seed production.

Our hybridisation experiment was restricted to using orchids that were already in cultivation with known provenance, i.e., they were not wild-collected. The next step is to expand hybridisation to a wider range of parents by collecting pollinia from wild orchids. This would provide greater scope for integrating more desirable traits, such as selecting parents on the basis of their volatile chemical profile, vigorous above ground growth, or proliferation of their underground tubers (Şavşatlı, 2023).

Both wild and artificial plant hybrids have demonstrated utility in conservation and sustainable trade. For example, in conservation of genetic diversity, Butternut (Juglans cinerea) has declined across its native range in eastern North America due to butternut canker (Ophiognomonia clavigignenti-juglandacaerum), a fungal pathogen. Buartnut (J. x bixbyi) is a naturally occurring hybrid with introduced Japanese walnut (J. ailantifolia), which tolerates this infection. A framework for conserving butternut proposes backcrossing buartnut with butternut to enhance butternut canker-resistance while retaining native germplasm (Pike et al., 2021). An example of sustainable trade is Tahitian vanilla (Vanilla × tahitensis), an edible orchid hybrid with high value, noted for its floral aroma, due to the anisyl alcohol content of its seeds. Its parents are V. planifolia and V. odorata (Lubinsky et al., 2008; Nielsen-Massey, 2024; Pérez-Silva et al., 2021).

Previously, attempts were undertaken to make salep trade more sustainable by creating synthetic products packaged in sachets made with corn starch and artificial sweeteners to thicken milk and give it a smooth, creamy texture just like real salep. Such artificial salep products are much cheaper to produce than collecting, drying, and grinding real orchid tubers into powder and, therefore, also much more affordable. It was hoped that artificial salep would reduce the pressure on wild-collected orchids because of its low costs for both producers and consumers. This did not happen, however, because of rising global demand for authentic food products. We developed a method to make salep trade more sustainable by using real orchids. Although rather than collecting them from the wild, we propose to create and grow artificially made sterile hybrids.

As previously noted, the higher cost of orchid tubers from cultivated plants compared to wild collected plants is a barrier (Masters et al., 2022). Analysis of economic viability is a stage for subsequent investigation. Using tissue cultivation for propagation could reduce the cost of purchasing juvenile plants for cultivation. Similarly hybrids producing 2 or more new tubers would sustain replanting from cultivated stock and having tubers to sell. To outcompete harvesting of wild orchids, additional costs of growing such hybrids and harvesting the tubers should be kept at a minimum as well. This could, for instance, be done by combining outdoor cultivation with other crops in mixed farming where local wild orchid populations have already gone extinct.

Focusing on technical feasibility of an identifiably cultivated orchid, in this case via hybridisation, was the purpose of this research. However, we recognize that collecting tubers from wild orchids for salep trade is a component of rural income (Ertuğ, 2000). Cultivation schemes can ensure that income received benefits community members and is not only accessible to landowners. An example of returning benefits to the community from formerly wild collected orchids is the cultivation of Cymbidium orchids supported by the government of Bhutan (Masters, 2015). One of the resolutions passed at the International Orchid Conservation Congress 2016 was that propagation of threatened orchids by small and local enterprises should be supported for sustainable production (Gale et al., 2018).