Knowing the level of genetic diversity and structure in marginal plant populations is essential for managing their genetic resources. This is particularly important for rare scattered tree species, such as Sorbus torminalis (L.) Crantz. We investigated the genetic diversity and its spatial distribution in peripheral populations of S. torminalis. As the species is known to reproduce vegetatively, we also evaluated clonal structure within populations. Using 13 nuclear microsatellite loci designed in two multiplexes, we genotyped 172 individuals revealing the existence of 100 distinct genotypes. Number of ramets per genotype was variable across populations with an average of 1.72. Examples of somaclonal variation at particular loci were detected. Measures of genetic diversity of the total sample were relatively high (mean allelic richness AR = 10.293; expected heterozygosity He = 0.756), as compared to other S. torminalis populations. We noticed a slightly negative inbreeding coefficient (FIS = -0.029) indicating a small excess of heterozygotes, which is typical for self-incompatible plants. Genetic differentiation among populations was low (FST = 0.048), but Bayesian clustering methods revealed the existence of three distinct genetic clusters only in part related to population structure. Significant spatial genetic structure within populations was also detected (Sp = 0.0125) indicating fine-scale pattern of isolation by distance. Our study demonstrated that peripheral populations of S. torminalis may exhibit relatively high levels of genetic diversity despite the existence of vegetative propagation. Nevertheless, if the studied or similar populations are expected to be utilized as seed sources for ex-situ or in-situ conservation purposes, the existence of clonal structure and spatial genetic structure must be taken into account in order to avoid excessive sampling of the same or closely related genets.

Several genera formerly contained within the genus Sorbus L. sensu lato have been proposed as separate taxa, including Aria, Chamaemespilus and Torminalis. However, molecular evidence for such distinctions are rather scarce. We assembled the complete chloroplast genome of Sorbus aucuparia, another representative of Sorbus s.s., and performed detailed comparisons with the available genomes of Aria edulis, Chamaemespilus alpina and Torminalis glaberrima. Additionally, using 110 complete chloroplast genomes of the Maleae representatives, we constructed the phylogenetic tree of the tribe using Maximum Likelihood methods. The chloroplast genome of S. aucuparia was found to be similar to other species within Maleae. The phylogenetic tree of the Maleae tribe indicated that A. edulis, C. alpina and T. glaberrima formed a concise group belonging to a different clade (related to Malus) than the one including Sorbus s.s. (related to Pyrus). However, Aria and Chamaemespilus appeared to be more closely related to each other than to Torminalis. Our results provide additional support for considering Aria, Chamaemespilus and Torminalis as separate genera different from Sorbus s.s.

Distribution of genetic diversity among and within plant populations may depend on the mating system and the mechanisms underlying the efficiency of pollen and seed dispersal. In self-incompatible species, negative frequency-dependent selection acting on the self-incompatibility locus is expected to decrease intensity of spatial genetic structure (SGS) and to reduce population differentiation. We investigated two populations (peripheral and more central) of wild service tree (Sorbus torminalis (L.) Crantz), a self-incompatible, scattered tree species to test the differences in population differentiation and spatial genetic structure assessed at the self-incompatibility locus and neutral nuclear microsatellites. Although, both populations exhibited similar levels of genetic diversity regardless of the marker type, significant differentiation was noticed. Differences between F ST and R ST suggested that in the case of microsatellites both mutations and drift were responsible for the observed differentiation level, but in the case of the S-RNase locus drift played a major role. Microsatellites indicated a similar and significant level of spatial genetic structure in both populations; however, at the S-RNase locus significant spatial genetic structure was found only in the fragmented population located at the north-eastern species range limits. Differences in SGS between the populations detected at the self-incompatibility locus were attributed mainly to the differences in fragmentation and population history.