Koen De Hert

• Premise of the study: In food‐deceptive orchid species, postmating reproductive barriers (fruit set and embryo mortality) have been shown to be more important for reproductive isolation than premating barriers (pollinator isolation). However, currently there is very little knowledge about whether germination failure acts as a reproductive barrier in hybridizing orchid species.• Methods: In this study, we investigated germination and protocorm development of pure and hybrid seeds of three species of the orchid genus Dactylorhiza. To test the hypothesis that germination failure contributed to total reproductive isolation, reproductive barriers based on germination were combined with already available data on early acting barriers (fruit set and embryo mortality) to calculate the relative and absolute contributions of these barriers to reproductive isolation.• Key Results: Protocorms were formed in all crosses, indicating that both hybrid and pure seeds were able to germinate and grow into protocorms. Also, the number of protocorms per seed packet was not significantly different between hybrid and pure seeds. High fruit set, high seed viability, and substantial seed germination resulted in very low reproductive isolation (average RI = 0.05). In two of six interspecific crosses, hybrids performed even better than the intraspecific crosses.• Conclusions: Very weak postmating reproductive barriers were observed between our study species and may explain the frequent occurrence of first‐generation hybrids in mixed Dactylorhiza populations. Germination failure, which is regarded as one of the most important bottlenecks in the orchid life cycle, was not important for reproductive isolation.

The potential for gene exchange between species with different ploidy levels has long been recognized, but only a few studies have tested this hypothesis in situ and most of them focused on not more than two co-occurring species. In this study, we examined hybridization patterns in two sites containing three species of the genus Dactylorhiza (diploid D. incarnata and D. fuchsii and their allotetraploid derivative D. praetermissa). To compare the strength of reproductive barriers between diploid species, and between diploid and tetraploid species, crossing experiments were combined with morphometric and molecular analyses using amplified fragment length polymorphism markers, whereas flow cytometric analyses were used to verify the hybrid origin of putative hybrids. In both sites, extensive hybridization was observed, indicating that gene flow between species is possible within the investigated populations. Bayesian assignment analyses indicated that the majority of hybrids were F₁ hybrids, but in some cases triple hybrids (hybrids with three species as parents) were observed, suggesting secondary gene flow. Crossing experiments showed that only crosses between pure species yielded a high percentage of viable seeds. When hybrids were involved as either pollen-receptor or pollen-donor, almost no viable seeds were formed, indicating strong post-zygotic reproductive isolation and high sterility. Strong post-mating reproductive barriers prevent local breakdown of species boundaries in Dactylorhiza despite frequent hybridization between parental species. However, the presence of triple hybrids indicates that in some cases hybridization may extend the F₁ generation.

• Premise of the study: Although the potential for gene flow between species with large differences in chromosome numbers has long been recognized, only few studies have thoroughly investigated in situ hybridization across taxa with different ploidy levels. We combined morphological, cytological, and genetic marker data with pollination experiments to investigate the degree, direction, and spatial pattern of hybridization between the diploid Dactylorhiza incarnata and its tetraploid derivative, D. praetermissa.• Methods: To identify hybrids, 169 individuals were genotyped using AFLPs and morphologically characterized. Individuals were clustered on the basis of their AFLP profile using the program Structure. To reduce the dimensionality of the plant‐trait matrix, PCA was applied. The origin of suspected hybrid individuals was verified using flow cytometry. An AMOVA and spatial autocorrelation analysis were used to indirectly infer the extent of gene flow.• Key results: Only five individuals were regarded as putative hybrids on the basis of the AFLP data; all had been assigned to the D. praetermissa morphotype. Only one had deviating DNA content and was presumably a triploid. High Φ_ST values between different subpopulations and significant spatial genetic structure were observed, suggesting localized gene flow.• Conclusions: Using combined data to study hybridization between D. incarnata and D. praetermissa, very few unequivocal hybrids were observed. We propose several non‐mutually exclusive explanations. Localized pollen flow, in combination with different microhabitat preferences, is probably one of the reasons for the low frequency of hybrids. Also, the triploid first‐generation hybrids may experience difficulties in successful establishment, as a result of genic incompatibilities.