Miscanthus is a genus of high‐yielding perennial rhizomatous grasses with C4 photosynthesis. Extensive field trials of Miscanthus spp. biomass production in Europe during the past decade have shown several limitations of the most widely planted clone, M. × giganteus Greef et Deu. A 3‐yr study was conducted at five sites in Europe (Sweden, Denmark, England, Germany, and Portugal) to evaluate adaptation and biomass production potential of four acquisitions of M. × giganteus (No. 1–4) and 11 other genotypes, including M. sacchariflorus (Maxim.) Benth. (No. 5), M. sinensis Andersson (No. 11–15), and hybrids (No. 6–10). At each site, three randomized blocks containing a 5‐ by 5‐m plot of each genotype were established (except in Portugal where there were two blocks) with micropropagated plants at 2 plants m−2. In Sweden and Denmark, only M. sinensis and its hybrids satisfactorily survived the first winter following planting. Mean annual yields across all sites for all surviving genotypes increased each year from 2 t ha−1 dry matter following the first year of growth to 9 and 18 t ha−1 following the second and third year, respectively. Highest autumn yields at sites in Sweden, Denmark, England, and Germany were 24.7 (M. sinensis hybrid no. 8), 18.2 (M. sinensis hybrid no. 10), 18.7 (M. × giganteus no. 3), and 29.1 t ha−1 (M. × giganteus no. 4), respectively. In Portugal, where irrigation was used, the top‐yielding genotype produced 40.9 t ha−1 dry matter (M. sinensis hybrid no. 7). Highest‐yielding genotypes in Sweden and Denmark were among the lowest yielding in Portugal and Germany, demonstrating strong genotype × environment interactions.

This paper describes the complete findings of the EU-funded research project OPTIMISC, which investigated methods to optimize the production and use of miscanthus biomass. Miscanthus bioenergy and bioproduct chains were investigated by trialing 15 diverse germplasm types in a range of climatic and soil environments across central Europe, Ukraine, Russia, and China. The abiotic stress tolerances of a wider panel of 100 germplasm types to drought, salinity, and low temperatures were measured in the laboratory and a field trial in Belgium. A small selection of germplasm types was evaluated for performance in grasslands on marginal sites in Germany and the UK. The growth traits underlying biomass yield and quality were measured to improve regional estimates of feedstock availability. Several potential high-value bioproducts were identified. The combined results provide recommendations to policymakers, growers and industry. The major technical advances in miscanthus production achieved by OPTIMISC include: (1) demonstration that novel hybrids can out-yield the standard commercially grown genotype Miscanthus x giganteus; (2) characterization of the interactions of physiological growth responses with environmental variation within and between sites; (3) quantification of biomass-quality-relevant traits; (4) abiotic stress tolerances of miscanthus genotypes; (5) selections suitable for production on marginal land; (6) field establishment methods for seeds using plugs; (7) evaluation of harvesting methods; and (8) quantification of energy used in densification (pellet) technologies with a range of hybrids with differences in stem wall properties. End-user needs were addressed by demonstrating the potential of optimizing miscanthus biomass composition for the production of ethanol and biogas as well as for combustion. The costs and life-cycle assessment of seven miscanthus-based value chains, including small- and large-scale heat and power, ethanol, biogas, and insulation material production, revealed GHG-emission- and fossil-energy-saving potentials of up to 30.6 t CO2eq C ha−1y−1 and 429 GJ ha−1y−1, respectively. Transport distance was identified as an important cost factor. Negative carbon mitigation costs of –78€ t−1 CO2eq C were recorded for local biomass use. The OPTIMISC results demonstrate the potential of miscanthus as a crop for marginal sites and provide information and technologies for the commercial implementation of miscanthus-based value chains.

The reasons for these requirements can be summarized as follows. Biomass with moisture contents below Miscanthus spp. are high-yielding perennial C 4 grasses, native to 200 to 250 g kg Ϫ1 fresh matter can be stored safely Asia, that are being investigated in Europe as potential biofuels. Production of economically viable solid biofuel must combine high without the danger of self ignition (Clausen, 1994) and biomass yields with good combustion qualities. Good biomass com-burns more efficiently while ash lowers the heating value bustion quality depends on minimizing moisture, ash, K, chloride, N, of the biomass and causes slagging of the boiler heat and S. To this end, field trials at five sites in Europe from Sweden exchangers (Hartmann et al., 1999). High levels of K to Portugal were planted with 15 different genotypes including M. ϫ are undesirable because it decreases the ash melting giganteus, M. sacchariflorus, M. sinensis, and newly bred M. sinensis point, but critical levels will depend on combustion techhybrids. Yield and combustion quality at an autumn and a late winter/ nique. Chloride can lead to corrosion through reaction early spring harvest were determined in the third year after planting with water to form HCl or with K to form gaseous when the stands had reached maturity. As expected, delaying the KCl, both of which are corrosive and reduce boiler life harvest by three to four months improved the combustion quality of (Baumbach et al., 1997). Furthermore, high chloride all genotypes by reducing ash (from 40 to 25 g kg Ϫ1 dry matter), K (from 9 to 4 g kg Ϫ1 dry matter), chloride (from 4 to 1 g kg Ϫ1 dry concentrations can lead to emissions of dioxine and matter), N (from 5 to 4 g kg Ϫ1 dry matter), and moisture (from furane (Siegle and Spliethoff, 1999). Nitrogen concen-564 to 291 g kg Ϫ1 fresh matter). However, the delayed harvest also trations in biofuels need to be as low as possible to decreased mean biomass yields from 17 to 14 t ha Ϫ1 . There is a strong minimize fertilizer off-takes and to reduce emissions interaction among yield, quality, and site growing conditions. Results of NO x during combustion. To avoid SO 2 emissions, show that in northern regions of Europe, M. sinensis hybrids can be biomass S concentrations also need to be as low as recommended for high yields (yielding up to 25 t ha Ϫ1 ), but M. sinensis possible. (nonhybrid) genotypes have higher combustion qualities. In mid-and To date, most research on Miscanthus sp. as an energy south Europe, M. ϫ giganteus (yielding up to 38 t ha Ϫ1 ) or specific crop has concentrated on maximizing the yield of a high-yielding M. sinensis hybrids (yielding up to 41 t ha Ϫ1 ) are more genotypes selected, there were four acquisitions of M. ϫ gigan-

Field trials in Europe with Miscanthus over the past 25 years have demonstrated that interspecies hybrids such as M. 9 giganteus (M 9 g) combine both high yield potentials and low inputs in a wide range of soils and climates. Miscanthus hybrids are expected to play a major role in the provision of perennial lignocellulosic biomass across much of Europe as part of a lower carbon economy. However, even with favourable policies in some European countries, uptake has been slow. M 9 g, as a sterile clone, can only be propagated vegetatively, which leads to high establishment costs and low multiplication rates. Consequently, a decade ago, a strategic decision to develop rapidly multiplied seeded hybrids was taken. To make progress on this goal, we have (1) harnessed Correspondence: John Clifton-

Genetic improvement through breeding is one of the key approaches to increasing biomass supply. This paper documents the breeding progress to date for four perennial biomass crops (PBCs) that have high output–input energy ratios: namely Panicum virgatum (switchgrass), species of the genera Miscanthus (miscanthus), Salix (willow) and Populus (poplar). For each crop, we report on the size of germplasm collections, the efforts to date to phenotype and genotype, the diversity available for breeding and on the scale of breeding work as indicated by number of attempted crosses. We also report on the development of faster and more precise breeding using molecular breeding techniques. Poplar is the model tree for genetic studies and is furthest ahead in terms of biological knowledge and genetic resources. Linkage maps, transgenesis and genome editing methods are now being used in commercially focused poplar breeding. These are in development in switchgrass, miscanthus and willow generating large genetic and phenotypic data sets requiring concomitant efforts in informatics to create summaries that can be accessed and used by practical breeders. Cultivars of switchgrass and miscanthus can be seed‐based synthetic populations, semihybrids or clones. Willow and poplar cultivars are commercially deployed as clones. At local and regional level, the most advanced cultivars in each crop are at technology readiness levels which could be scaled to planting rates of thousands of hectares per year in about 5 years with existing commercial developers. Investment in further development of better cultivars is subject to current market failure and the long breeding cycles. We conclude that sustained public investment in breeding plays a key role in delivering future mass‐scale deployment of PBCs.