Twelve loci with the best primer scores were selected from a shotgun 454-sequenced genomic library to investigate levels of genetic diversity at microsatellite loci in L. subulatum (Table 1).

The 12 polymorphic microsatellite loci detected a total of 80 different SSR alleles in the 35 individuals analysed of L. subulatum. The number of alleles ranged from a minimum of two alleles for locus Lsub09 to a maximum of 12 alleles for loci Lsub02 and Lsub04 and the mean number of alleles per locus was of 6.67 (Table 2). Observed heterozygosities ranged from H_O = 0.086 (locus Lsub09) to H_O = 0.914 (locus Lsub03) and expected heterozygosities ranged from H_E = 0.083 (locus Lsub09) to H_E = 0.881 (locus Lsub04). None of the 12 loci showed significant heterozygote deficiency, with an overall non-significant inbreeding coefficient value of F_IS = −0.030 (Table 2).

All 12 microsatellite loci were transferable to all assayed species (Table 2). All loci were polymorphic in the Iberian endemic gypsophyte, Lepidium cardamines (Sect. Dileptium). 8 and 10 of the 12 loci were polymorphic in the high mountain, Iberian, narrow endemic Lepidium stylatum (Sect. Lepia (Desv.) DC.), and in the widespread ruderal, L. graminifolium (Sect. Lepidium), respectively. However, in the latter species individual genotypes could not be reliably scored from the amplification patterns with individuals showing up to five amplified bands per individual (Table 2). Ten and eight loci were polymorphic in C. draba and C. didymus respectively, with amplification patterns revealing up to six bands per individual in both taxa (Table 2). Thus, the multiple banding patterns observed at each locus supported the polyploid status of Lepidium graminifolium (2n = 6x = 48 [2]), C. draba (2n = 8x = 64 [6]) and C. didymus (2n = 4x = 32).

In the analysed polyploid species banding patterns were consistent with the aforementioned ploidy levels. Up to four bands per individual were amplified in the tetraploid C. didymus, and up to six bands were amplified in the octoploid C. draba (Table 2). In this latter species number of bands did not reach the expected maximum of eight bands per individual, most likely because the loci were not variable enough in the analysed population. A similar result was obtained in the hexaploid L. gramminifolium with up to five bands amplified per individual. Interestingly, in this species six out of 10 polymorphic loci showed fixed heterozygous profiles with individuals showing consistently a minimum of two amplified bands. This may indicate an allopolyploid hybrid origin for the species [7,8].

Five hundred nanograms of total DNA were extracted from silica-gel dried young leaves of a pool of individuals using the DNeasy plant minikit (Qiagen, Barcelona, Spain) following the manufacturer’s instructions, and eluted in 130 μL TE. This DNA was used to construct 454 genomic libraries by the sequencing service from the University of Valencia (SCSIE, Spain) and to conduct shotgun sequencing on a GS Junior 454 sequencer (Roche, Barcelona, Spain). We obtained 119,732 reads with an average read length of 465.61 bp and a total amount of 55,748,981 bases that were trimmed of adaptor and low-quality regions and assembled into contigs using GS De Novo Assembler implemented in Newbler 2.5p1 (Roche, Madrid, Spain), using default parameters (e.g., 40 bases minimum overlap and 100 bases minimum contig length). Generated contigs and unique reads not assigned to contigs were subjected to BLAST analysis and those matching organellar (chloroplast or mitochondria) sequences were discarded.

We screened all 8305 unique reads and non organellar contigs with iQDD v. 1.3.0.0 software [9]. We set the script to identify all possible di-, tri, tetra-, penta- and hexanucleotide repeats with a minimum of 4 repeat units, and compound repeats. Mononucleotide repeats were not considered. A total of 969 reads containing microsatellites were found in the 8305 screened reads, consisting of 623 dinucleotide, 143 trinucleotide, three tetranuleotides, one pentanucleotide, one hexanucleotide and 198 compound repeats. The program was allowed for direct primer design using PRIMER3 [10] by searching for microsatellite repeats and primer annealing sites in the flanking regions. After discarding reads with too short flanking sequences, primers were successfully designed for 652 reads. However, 405 of these (62.12%) corresponded to AT/TA repeats and were not considered because of their self-complementarity within DNA strains which could make them are more prone to amplification artefacts [11]. Of the 247 remaining loci, 21 dinucleotide and 17 trinucleotide microsatellite loci had more than seven repeat units. Twelve loci out of the 38 showing the best primer scores were selected for PCR amplification and as all of them produced clear amplicons of the expected size in 2% agarose gels they were subsequently selected for analysis on automated sequencers (Table 1). Forward primers were labelled with 6FAM, NED, PET or VIC fluorescent dyes from Applied Biosystems for automated electrophoresis (Table 1). PCR amplifications were performed in a 20 μL mix containing 1× Taq Buffer (Biotools, Madrid, Spain), 2 mM MgCl₂, 0.4 mM of each dNTP, 5 pmol each of the labelled (forward) and unlabelled (reverse) primers, 1 U of Taq polymerase (Biotools) and 20 ng of template DNA. The PCR program consisted of one step of 4 min at 94 °C followed by 39 cycles each of 1 min at 94 °C, 1 min at 55 °C for annealing, and 1 min at 72 °C, and a final extension step of 7 min at 72 °C. The products were run on an ABI 3730XL automated sequencer (Applied Biosystems) using LIZ500 as the internal lane size standard, and the amplified fragment lengths were assigned to allelic sizes with GENEMARKER v. 1.85 software (SoftGenetics, State College, PA, USA). After an initial screening of individuals, all 12 loci showed consistent amplification patterns and polymorphisms and were subsequently used for genotyping the entire set of samples.

Genotypic data were obtained for 35 individuals from one population of L. subulatum (Table 3) for 12 microsatellite loci (Tables 1 and 2).

Number of alleles (N_A), observed (H_O) and unbiased expected (H_E) heterozygosities [12] were calculated with GENETIX v. 4.05 [13]. Inbreeding coefficients (F_IS) and deviations from Hardy-Weinberg equilibrium and linkage disequilibrium between pairs of microsatellite loci, using 1000 permutations, were calculated using GENEPOP v.4.1.4 software [14]. None of the 66 pairwise comparisons between loci showed significant linkage disequilibrium (p < 0.05).

Cross-species transferability was assayed in one population each of three other species of Lepidium (L. cardamines L., L. graminifolium L., and L. stylatum Lag. & Rodr., Table 3), and one species each of the closely related genera Cardaria Desv. (C. draba L.), and Coronopus Zinn (C. didymus L., Table 3).