Development of genetic markers from whole-genome shotgun sequences of Balsamocarpon brevifolium. Balsamocarpon brevifolium ssr marker

The development of species-specific molecular markers became relatively easy after the advent of next-generation sequencing, as obtaining genome information is no longer very expensive. However, low genomic coverage of the next-generation sequence reads can result in assembly artefacts and therefore reduce the reliability of the obtained marker loci. We here show for Balsamocarpon brevifolium, an endangered endemic shrub of the Chilean Atacama Desert, that an additional step of mapping sequence reads back to the initial sequence assembly can considerably improve the basis for marker development. Information regarding the population structure of B. brevifolium is extremely limited, as population genetic molecular markers are absent for this species. To assemble the complete plastid genome and to characterize microsatellite (SSR) loci, we carried out low-coverage shotgun sequencing (2x250 bp) of genomic DNA on the Illumina MiSeq platform. Out of five million quality-filtered reads we obtained 551,497 contigs (N50 = 375 bp). Using contigs with a very high coverage (500-2000-fold) we were able to assemble the plastid genome of B. brevifolium that is a source of maternally inherited genetic markers. In the nuclear genome we identified 6,042 loci with potentially useful microsatellite motifs (83.1% di- and 13.1% tri-nucleotide repeats). Only 2% out of the initially tested 135 tri-nucleotide loci provided reliable amplicons. Back mapping of sequence reads to our initial assembly revealed that the high rate of unsuccessful amplifications was caused by erroneously assembled contigs and/or heterozygosity of the loci at the primer binding sites. When we used loci that were initially confirmed by mapping reads back to the assembled contigs 75% of the tested loci could be reliably amplified. This highlights the need of additional control steps such as back mapping especially when dealing with low genomic coverage sequencing. We report here 16 unlinked biallelic SSR markers, of which the detected variation ranges between two and 11 alleles per locus (average = 6). Their utility was tested in ten individuals suggesting their suitability for population genetic analyses in B. brevifolium.