BACKGROUND AND AIMS:The advent of molecular breeding is advocated to improve the productivity and sustainability of second-generation bioenergy crops. Advanced molecular breeding in bioenergy crops relies on the ability to massively sample the genetic diversity. Genotyping-by-sequencing has become a widely adopted method for cost-effective genotyping. It basically requires no initial investment for design as compared with array-based platforms which have been shown to offer very robust assays. The latter, however, has the drawback of being limited to analyse only the genetic diversity accounted during selection of a set of polymorphisms and design of the assay. In contrast, genotyping-by-sequencing with random sampling of genomic loci via restriction enzymes or random priming has been shown to be fast and convenient but lacks the ability to target specific regions of the genome and to maintain high reproducibility across laboratories. METHODS:Here we present a first adoption of single-primer enrichment technology (SPET) which provides a highly efficient and scalable system to obtain targeted sequence-based large genotyping data sets, bridging the gaps between array-based systems and traditional sequencing-based protocols. To fully explore SPET performance, we conducted a benchmark study in ten Zea mays lines and a large-scale study of a natural black poplar population of 540 individuals with the aim of discovering polymorphisms associated with biomass-related traits. KEY RESULTS:Our results showed the ability of this technology to provide dense genotype information on a customized panel of selected polymorphisms, while yielding hundreds of thousands of untargeted variable sites. This provided an ideal resource for association analysis of natural populations harbouring unexplored allelic diversities and structure such as in black poplar. CONCLUSION:The improvement of sequencing throughput and the development of efficient library preparation methods has made it feasible to carry out targeted genotyping-by-sequencing experiments cost-competitively with either random complexity reduction systems or traditional array-based platforms, while maintaining the key advantages of both technologies.