Premise of the Research. Vascular plants depend on sexual recombination for generating new genetic variability to meet environmental needs. Nevertheless, members of the early-divergent fern genus Botrychium (Ophioglossaceae) typically maintain gametophytic selfing and show strong inbreeding within populations. To explain this evolutionary anomaly, the existence of previous or current but undiscovered outcrossing, genetically rich, precursors of the existing genetically depauperate taxa has been hypothesized. 

Methodology. Using allele expression at thirteen independently assorting enzyme loci, we compared allelic diversity and levels of heterozygosity in 471 Botrychium lunaria individuals across sixteen populations in the Alps and Jura Mountains of Switzerland. We examined habitat characteristics influencing mating systems and investigated population genetic structure based on a discriminant analysis of principal components and a graph-theoretic framework. We tested the pattern of isolation by distance and explored the biological processes favoring or limiting spore dispersal at a regional scale.

Pivotal Results. We found high genetic diversity within and among populations, and similar observed (H_O) and expected heterozygosity (H_S) across loci (H_O = 0.020-0.450 and H_S = 0.040-0.590, respectively). Estimates of H_O(0.050-0.272) were lower than H_S (0.144-0.305) across individuals, indicating a weak deficit in heterozygotes. Mean values of departure (D) from expected heterozygosity were close to what would be expected with random mating (D_L = 0.061 and D_P = 0.059 across loci and populations, respectively). The mean inbreeding coefficient was low (F_IS = 0.247) and the overall genetic differentiation was moderate (global F_ST = 0.083).

Conclusions. Together, the discriminant analysis of principal components and the Population Graph revealed a complex population genetic structure and a weak genetic differentiation with substantial gene flow among most populations. One population (Arosa), sampled from a site recently released from permafrost, demonstrated probable founder effect not correlated with geographical isolation. The topology of the Population Graph identified three subgroups with several key populations (BGU, CHL, GSB, SBE, and VRO) that maintained genetic connectivity of populations from distant locations.

Allozyme data were generated following the procedures of Stensvold & Farrar (2017). As soon as possible after collection, 1 cm of tissue was cut from the base of the leaf stalk and crushed in a buffer solution containing phosphate-polyvinylpyrrolidone (Cronn et al. 1997). The grindate was stored for up to 12 months in microcentrifuge tubes at –80°C until electrophoresis was conducted. Enzyme electrophoresis was performed in horizontal 12% starch gel using the three buffer systems 7, 9, and 11 from Soltis et al. (1983) to reveal allelic variants at 22 loci of ten enzyme systems by differentiating gel migration patterns. Thus, loci of those enzyme systems were resolved for aspartate aminotransferase (AAT) and triose-phosphate isomerase (TPI) with buffer system 7, for malate dehydrogenase (MDH), phosphoglucomutase (PGM), phosphogluconate dehydrogenase (PGD), and phospho-glucoisomerase (PGI) with buffer system 9 and, for aconitase (ACN), diaphorase (DIA), isocitrate dehydrogenase (IDH), and shikimate dehydrogenase (SKDH) with buffer system 11. The stain recipes used followed Soltis et al. (1983).

Allele letter codes (i.e., scoring) were assigned with reference to alleles previously determined for Botrychium taxa of the B. lunaria complex by Stensvold & Farrar (2017). Consistency of scores was checked by multiple runs of the same specimens when necessary. Allele variants for each locus were numbered, assigning the lowest number to the fastest (most anodal) migrating allele. In analyses we removed rare alleles (population frequencies <0.05) to avoid artificial inflation of polymorphism estimates due to experimental error as recommended by Hartl & Clark (1998) and excluded uninformative loci that were monomorphic across all populations. Loci with missing scores in more than 30% of samples were excluded.