Résumé
Patterns of polyploid evolution in the taxonomically controversial Dactylorhiza incarnata/maculata groups were inferred genetically by analyzing 399 individuals from 177 localities for (1) four polymorphic plastid regions yielding aggregate haplotypes and (2) nuclear ribosomal ITS allele frequencies. Concordance between patterns observed in distributions of plastid haplotypes and ITS alleles renders ancestral polymorphism an unlikely cause of genetic variation in diploids and allopolyploids. Combining the degree of concerted evolution in ITS alleles (thought to reflect gene conversion) with inferred parentage provides support for a quadripartite classification of western European allopolyploid dactylorchids according to their respective parentage and relative dates of origin. The older allotetraploids that generally exhibit only one parental ITS allele can be divided into those derived via hybridization between the divergent complexes we now call D. incarnata s.l. and D. fuchsii (e.g., D. majalis) and those derived via hybridization between D. incarnata s.l. and D. maculata (e.g., D. elata). Similarly, the younger allotetraploids that maintain evidence of both parental ITS alleles can be divided into those derived from hybridization between D. incarnata s.l. and D. fuchsii, or perhaps in some cases a diploid species resembling D. saccifera (e.g., D. praetermissa, D. purpurella, D. traunsteineri s.l., D. baltica), and those derived from hybridization between the D. incarnata s.l. and D. maculata groups (e.g., D. occidentalis, D. sphagnicola). Older allotetraploids are inferred to have passed through glacially induced migration bottlenecks in southern Eurasia, whereas at least some younger allotetraploids now occupying northern Europe are inferred to have originated post-glacially and remain sympatric with their parents, a scenario that is largely in agreement with the morphology and ecology of these allotetraploids. ITS conversion is in most cases biased toward the maternal parent, eventually obscuring evidence ofthe original allopolyploidization event because plastid haplotypes also reflect the maternal contribution. Gene flow appears unexpectedly low among allotetraploids relative to diploids, whereas several mechanisms may assist the gene flow observed across ploidy levels. There is good concordance between (1) the genetically delimited species that are required to accurately represent the inferred evolutionary events and processes and (2) morphologically based species recognized in certain moderately conservative morphological classifications previously proposed for the genus. Further research will seek to improve sampling, especially in eastern Eurasia, and to develop more sensitive markers for distinguishing different lineages within (1) the remarkably genetically uniform D. incarnata group (diploids) and (2) locally differentiated populations of (in some cases unnamed) allotetraploids.