Variation in chilling tolerance for photosynthesis and leaf extension growth among genotypes related to the C4 grass Miscanthus ×giganteus

Summary Tetraploid M. sacchariflorus were identified as having equal chilling tolerance to M. ×giganteus ‘Illinois’, but M. sinensis were inferior. Heterosis did not explain the chilling tolerance of M. ×giganteus, but suboptimal ploidy reduced it.


Introduction
During the last two decades, Miscanthus ×gigantues (Mxg) has become an important biomass crop for European and US bioenergy initiatives (Scurlock, 1999;Clifton-Brown et al., 2004;Heaton et al., 2008;Somerville et al., 2010). By combining perenniality, rhizome function, and cold-tolerant C 4 photosynthesis, Mxg is highly light, nitrogen, and water use efficient, which contributes to its exceptionally high biomass productivity even in cool temperate climates (Lewandowski et al., 2000;Heaton et al., 2008;Long and Spence, 2013). While many C 4 plants are now known to survive cold conditions, Mxg appears exceptional in its ability to achieve the high productivity of C 4 photosynthesis in cold climates. For example, in southeastern England, Mxg shoot productivity exceeded the highest values achieved by intensively managed C 3 crops (Beale and Long, 1995). A comparative field study with a high-yielding maize cultivar (Illinois, USA) revealed that perenniality together with cold tolerance gave Mxg 59% more productivity than maize (Dohleman and Long, 2009). This was largely explained by the ability of Mxg to form photosynthetically competent leaves earlier in the year and maintain them later in the year. This allowed Mxg to capture 60% more solar energy than maize, and yet convert this into biomass with a similar efficiency, despite colder weather at the beginning and end of the growing season (Dohleman and Long, 2009).
A single genotype of Mxg collected from southern Honshu, Japan was initially introduced to Denmark in the 1930s and was then subsequently distributed throughout Europe and North America (Greef and Deuter, 1993;Hodkinson et al., 2002;Głowacka et al., 2014). This cultivar is a triploid sterile hybrid between diploid M. sinensis (Msi) and tetraploid M. sacchariflorus (Msa) (Greef and Deuter, 1993;Linde-Laursen, 1993;Hodkinson et al., 2002). Hodkinson and Renvoize (2001) defined the nothospecies, Miscanthus ×gigantues Greef & Deu. ex Hodkinson & Renvoize, as a hybrid between Msi and Msa, a designation that could include an infinite number of genotypes. However, almost all trials and production fields of Mxg appear to be from the single clone first introduced to Denmark, though limited information about new natural Mxg accessions and those produced by controlled crosses is beginning to appear (Nishiwaki et al., 2011;Jeżowski et al., 2011;Dwiyanti et al., 2013). Additionally, new genotypes of Mxg have recently been bred at the Univeristy of Illinois via controlled crosses (Chae et al., 2014). The first replicated trials of Mxg in the USA were undertaken in Illinois by Heaton et al. (2008), and this has since become known as the 'Illinois' clone. While the 'Illinois' clone has been shown to achieve high productivity and is being used in small-and mediumscale production operations, it does have limitations. First, a single clone is highly vulnerable to potential diseases and pests. Secondly, its origin in southern Honshu is far south of the northern limit of Miscanthus in eastern Asia. Therefore, it is likely that there is greater cold tolerance in the germplasm that would allow the development of cultivars for colder climates than that of England and Illinois or that would have a longer potential growing season within these climates. Thus, there is a strong need for breeding efforts to produce and evaluate new Mxg genotypes. Indeed since the 'Illinois' clone is a sterile triploid hybrid, finding potential parent lines with at least an equivalent chilling tolerance will be important for developing new hybrids adapted to temperate environments. Such genotypes would be useful not only for Miscanthus breeding but also for improving cold tolerance in sugarcane. Miscanthus has been hybridized with and subsequently backcrossed to sugarcane, facilitating introgression of disease tolerance genes (Chen et al., 1993). It would therefore seem likely that the same approach could be used to introgress cold tolerance from selected Miscanthus genotypes into sugarcane.
To date, most studies of chilling-tolerant photosynthesis in Miscanthus have focused on the single commercial genotype of Mxg (Beale et al., 1996;Naidu et al., 2003;Naidu and Long, 2004;Farage et al., 2006;Wang et al., 2008a, b). An exception was an analysis of the response of photosynthesis to chilling (12 °C for 12 h) for two Msi hybrids, one tetraploid Msa and one Mxg, by Purdy et al. (2013). Additionally, Clifton-Brown and Jones (1997) found that genotypic variation in leaf expansion at 5 °C relative to 20 °C shows a strong correspondence to final yield in Miscanthus.
Exposure to chilling temperatures (>0≤14 °C) has reversible as well as irreversible effects on C 4 photosynthesis. Chilling lowers the capacity for CO 2 assimilation and, in the absence of adequate alternative pathways for dissipating absorbed excitation energy, can lead to reversible photoinhibition or irreversible photooxidation (Long, 1983;Long et al., 1994;Allen and Ort, 2001). Mxg has been shown to avoid this damage by increasing photosynthetic capacity under chilling conditions (Wang et al., 2008b) and by increasing its capacity for non-photochemical quenching of absorbed light energy (Farage et al., 2006).
In Mxg this chilling tolerance might come from one or both parents, or be a result of hybrid vigour. It might also be expected to come from polyploidy, since this could amplify genes conferring chilling tolerance. The recent development of a hexaploid by chromosome doubling of the 'Illinois' clone, as well as the discovery of new triploid and diploid forms of Mxg, provide an opportunity to test these hypotheses. When considering variation in overall chilling tolerance among genotypes, variation for chilling tolerance of leaf extension growth must also be considered, since both leaf extension and photosynthesis must have similar chilling tolerance to provide a productivity advantage to a cultivar. In Miscanthus, leaf extension in low temperature is indicative of chilling tolerance (Clifton-Brown and Jones, 1997), and specific leaf area can correlate with CO 2 assimilation rate at low temperature (U. Jørgensen, personal communication).
The foregoing gives rise to a series of questions about genetic variation in relatives of Mxg that are investigated here. (i) What variation in chilling tolerance can be found in different genotypes of Mxg and in Mxg's parental species? (ii) Can chilling tolerance equal to or better than that of the Mxg 'Illinois' be found? (iii) From which of the parental Mxg species is chilling tolerance most probably inherited? (iv) Is there an association between increasing ploidy level and chilling tolerance? To address these questions, low temperature leaf elongation of Mxg 'Illinois' together with 50 additional Miscanthus genotypes belonging to seven species-ploidy groups were determined, and a subset were further evaluated for photosynthetic capacity during and after chilling conditions.

Plant material
A total of 51 Miscanthus accessions were studied (Table 1): three diploid Msa, 10 tetraploid Msa, 27 diploid Msi, five tripolid Mxg, three colchicine-induced polyploids (5x-6x) of Mxg 'Illinois' (Chae et al., 2013), one diploid Mxg, one M. oligostachyus 'Purpurascens', and a Miscanthus accession of unknown provenance named Miscanthus sp. 'PF1-7'. Most of the Msi genotypes were cultivars commercially available in the USA, and chosen based on spring and/ or autumn hardiness, and other features related to cold tolerance, namely early emergence, late flowering, and remaining green after autumn frost (C. Kaiser and E. Sacks, unpublished data). Most of  (Chae et al., 2014). Two sugarcane (Saccharum officinarum) cultivars and two Zea mays lines were included as controls. Clonal divisions of Miscanthus and sugarcane accessions as well as Z. mays seedlings were grown in square pots of 1 litre volume (mini-treepot # MT38; Stuewe & Sons, Tangent, OR, USA) containing a peat-, bark-, and perlite-based growing medium (Metro-Mix 900; Sun Gro Horticulture, Agawam, MA, USA). After planting or sowing, an all-purpose slow release fertilizer was added following the manufacturer's instructions (Osmocote Classic, 8-9 mo 13-13-13; Everris NA, Inc., Dublin, OH, USA), and additional iron added (ferrous sulphate heptahydrate; QC Corporation, Girardeau, MO, USA). Prior to transfer to controlled-environment cabinets, plants were grown in a controlled environment greenhouse at ~25 °C. Throughout, soil moisture content was maintained by watering to field capacity daily.

Leaf elongation
All 55 accessions (51 Miscanthus, two sugarcane, and two maize) were evaluated for leaf elongation under chilling temperatures, enabling those with the greatest and least chilling tolerance to be identified efficiently for further study. Plants were transferred to controlled-environment chambers (BioChambers GRC-40; Conviron, Winnipeg, Manitoba, Canada) with a 12 h day/12 h night cycle under 1000 μmol photons m -2 s -1 , a constant temperature of 25 °C, and relative humidity of 50%. After 10 d acclimation to the controlled-environment chamber, three out of six plants from each accession were selected at random, and transferred to a second identical controlledenvironment chamber, but with a 10 °C/5 °C day/night temperature. All other environmental conditions were unchanged. To avoid confounding genotype with any undetected variation, the position of each plant within the chamber was changed following a randomized design every second day. Data were collected on the uppermost leaf on the stem in which the ligule had not emerged from the sheath of the preceding leaf (i.e. growing leaves). The distance from the top edge of the pot to the tip of the chosen leaf was measured using a ruler. For each plant, leaf growth was calculated by subtracting the measurement on day 0 of the treatment from the measurements collected on subsequent days. In the 10 °C/5 °C treatment, developing leaves were measured every second day, and every day in the control.

Gas exchange and chlorophyll fluorescence
The high-throughput leaf elongation experiment served as a preliminary screen to select the most cold-tolerant genotypes for a more detailed but lower throughput assessment of photosynthetic response to chilling. A subset of 13 accessions, chosen from those which showed the greatest percentage of leaf elongation retained in chilling temperature relative to warm temperature and three negative controls (16 accessions in total; Fig. 1, grey-and yellow-marked accessions), was analysed for photosynthetic capacity under chilling conditions. Four plants of each accession were gown at 25 °C/20 °C (warm) in each of two controlled-environment chambers (Conviron PGR15; Controlled Environments) for 10 d, after which the temperature in both chambers was lowered to 10 °C/5 °C (chilling) day/night for 11 d, and then returned to 25 °C/20 °C. This period was chosen to mimic the type of chilling that might develop during spring after leaf emergence or expanded leaves in the autumn. Loss of efficiency during these periods greatly limits realization of the superior efficiency of C 4 relative to C 3 photosynthesis in temperate climates (Long and Spence, 2013). The position of plants within the chambers was changed, as described for the leaf elongation study. In both chambers a 14 h day/10 h night cycle with 1000 μmol photons m -2 s -1 and relative humidity of 65% was maintained. Leaf photosynthetic gas exchange and modulated chlorophyll fluorescence were measured in situ on the most recent fully expanded attached leaves, as judged by ligule emergence, with an open gas exchange system incorporating differential infrared CO 2 and water vapour analysers (LI-6400; LI-COR, Lincoln, NE, USA). In this system, the leaf was enclosed in a controlled-environment cuvette which tracked the light, temperature, and humidity in the controlled-environment chamber. Chlorophyll pulse amplitude modulated fluorescence was measured simultaneously with a fluorometer incorporated into the cuvette lid (LI-6400-40; LI-COR, Inc.). Measurements were conducted under ambient air (21% O 2 ) at 390 μmol mol -1 CO 2 concentration, 1000 μmol m -2 s -1 photon flux, and 65% relative humidity. Leaf temperature was maintained at the growth temperature for each accession and treatment. Actinic light was supplied by lightemitting diodes (90% red light, 630 nm; 10% blue light, 470 nm). To maximize the fluorescence emissions, the fluorometer parameters (e.g. flash intensity and duration) were adjusted and the multiphase protocol was used. These measurements were made in warm conditions (25 °C) just prior to the chilling treatment, immediately after the temperature was reduced to 10 °C (day 0), during each of the 11 d at 10 °C (except days 6, 8, and 10 when the measurements were not taken), and finally 1 d after transferring the plants back to 25 °C (12th day of the experiment-recovery). All measurements were taken during the day on light-adapted leaves until steady state (20-50 min). On day 0, in order to obtain measurements for each plant immediately after reducing the temperature to 10 °C, each pot was individually transferred from the 25 °C growing chamber to a growing chamber at 10 °C; after the leaves reached steady-state readings, gas exchange parameter were recorded. For each accession, four replicate plants were measured for each treatment. From these measurements, leaf net CO 2 uptake per unit leaf area (A), stomatal conductance to water vapour (g s ), intercellular CO 2 concentration (c i ), the quantum yield of photosystem II (Ф PSII ), and rate of whole chain electron transport (J) were calculated, as described previously (Bernacchi et al., 2003).

Data analysis
All statistical analyses were performed with SAS v. 9.3 (SAS Institute, Cary, NC, USA). Leaf elongation was regressed against time to determine the best-fitting equation to describe the rate of

Genotypic variation in leaf elongation
Mxg (3x) 'Illinois', the reference chilling-tolerant accession, showed a 9-fold decrease in leaf elongation rate from 22.9 mm d -1 at 25 ºC to 2.5 mm d -1 at 10 °C/5 °C (Fig. 1A,  B). Of the 50 Miscanthus genotypes analysed, two groups were identified, one with 17 accessions that were not significantly different from Mxg (3x) 'Illinois' and the other with 32 accessions that showed a significantly greater reduction in leaf extension growth (P≤0.1; Fig. 1B). Notably, one accession, Msa 'PF30154', significantly exceeded the leaf elongation of Mxg (3x) 'Illinois' by 1.5-fold when grown at chilling temperature (Fig. 1B). There was considerable variation between Miscanthus accessions, with Msa 'PF30154' showing 43-fold greater leaf extension growth at 10 °C/5 °C than the worst performing accession, diploid Msi 'Uryu' (Fig. 1B). The two cultivars of Z. mays and two accessions of S. officinarum showed some of the lowest extension rates at 10 °C/5 °C, but comparable with the worst performing Miscanthus accessions. However, at 25 °C, all four maize and sugarcane controls had significantly greater leaf elongation rates than Mxg (3x) 'Illinois'. Only two Miscanthus accessions matched these rates, Msa var. lutarioriparius 'PF30022' and Miscanthus sp. 'PF1-7'. However, Msa var. lutarioriparius showed only a 12-fold decline in leaf elongation in the chilling treatment, compared with >30-fold declines for the Saccharum accessions and the Z. mays cultivars. At 10 °C/5 °C, all 10 tested tetraploid Msa and two hexaploid Mxg were not significantly different from Mxg (3x) 'Illinois' (Fig. 1C). Two of the three diploid Msa were also not significantly different from Mxg (3x) 'Illinois'. In contrast to the Msa accessions, approximately two-thirds of the Msi accessions had significantly lower rates of leaf elongation at chilling temperatures than Mxg (3x) 'Illinois' (Figs 1, 2; Supplementary Fig. S1 available at JXB online).
Among four new triploid Mxg obtained from the cross, Msa (4x) 'Bluemel Giganteus'×Msi (2x) var. condensatus 'Cabaret', three of the progeny and their Msa parent were not significantly different from Mxg (3x) 'Illinois' for percentage of leaf elongation retained in chilling temperature relative to warm temperature (Fig. 1C). However, the Msi parent (P2 low; Fig. 1), was among the least cold-tolerant accessions studied (similar to sugarcane and maize) as assessed by leaf elongation.
Three main patterns of loss of photosynthetic capacity could be observed across the germplasm studied, over the 11 d following transfer to 10 ºC/5 ºC. These may be characterized as nearly full acclimatization, partial acclimatization, and failure to acclimatize ( Fig. 4; Supplementary Fig. S3 at JXB online). For Mxg (3x) 'Illinois', A sat and Ф PSII declined for the first 2-3 d at 10 ºC/5 ºC, followed by a recovery over the following days to stabilize at a level of 91-92% of the rates on initial transfer to 10 °C/5 ºC (Fig. 4A, C). Msa (4x) 'PF30153' was the only clone studied which showed a similar pattern of response and was not significantly different from Mxg (3x) 'Illinois' with respect to A sat at the end of the chilling treatment. Both Mxg (3x) 'Illinois' and Msa 'PF30153' recovered fully 1 d after transfer back to 25 °C, showing ~87% and 84%, respectively, of the pre-chilling A sat (Supplementary Fig. S3A). Additionally, two other tetraploid Msa, 'PF30151' and 'PF30154', did not differ significantly from Mxg (3x) 'Illinois' with respect to relative A sat at the end of the chilling treatment (A sat calculated as a percentage of initial values at day 0; Fig. 4). Compared with Mxg (3x) 'Illinois', all other Miscanthus clones showed a significantly lower A sat over the 11 d of chilling (Figs 3, 4). Although seven other genotypes, six Msa and Msi 'Nipon', showed a significantly lower A sat over the 11 d at 10 °C/5 °C compared with Mxg, on return to warm conditions they recovered an A sat after 1 d that was not significantly lower than the value for Mxg (3x) 'Illinois' (Fig. 3A). All other lines showed a significantly greater loss of photosynthetic capacity during the 11 d at 10 °C/5 °C and on return to warm conditions, although none showed as large a loss as the Z. mays and two Saccharum spp. lines (Figs 3A, 4; Supplementary Fig. S2A). Several lines showed an increase in stomatal conductance on initial transfer to 10 °C/5 °C ( Fig. 3; Supplementary Fig. S2). All lines showed an increase in c i /c a on transfer to 10 °C/5 °C, but only Z. mays and the two Saccharum spp. lines showed a significant increase in c i /c a over the 11 d of chilling compared with Mxg (3x) 'Illinois'. Both Saccharum spp. lines also showed an increase in c i /c a on return to 25 °C/20 °C relative to pre-chilling levels. In general, the other Mxg genotypes tested, including the hexaploid derived from the Mxg (3x) 'Illinois', were among the less chilling-tolerant lines, with respect to gas exchange measures (Figs 3, 4; Supplementary  Fig. S2). There was a close correlation between A sat and whole chain electron transport (J) calculated from Ф PSII both before and after the chilling treatment (Fig. 5). However, the A sat /J declined ~34% across the genotypes as a result of the chilling treatment (Fig. 5).

What variation in chilling tolerance can be found within Mxg?
Mxg (3x) 'Illinois', which is the same clone that has been widely trialled in Europe (Hodkinson et al., 2002;Głowacka et al., 2014), has exceptional chilling tolerance of photosynthesis and leaf elongation, underlying its high productivity in cool climates (Beale and Long, 1995;Beale et al., 1996;Wang et al., 2008 ;Dohleman and Long, 2009;Purdy et al., 2013;Arundale et al., 2014). However, the basis of this chilling tolerance remains unclear. One possibility is that the performance of this clone at chilling temperatures results from hybrid vigour rather than simple inheritance from its Msa and Msi parents. To test this hypothesis, new triploid hybrids were generated by crossing Msa (4x) 'Bluemel Giganteus' and Msi (2x) var. condensatus 'Cabaret'. It was known from previous experience that the 'Cabaret' parent was insufficiently hardy to over-winter undamaged in cold-temperate environments, such as in central Illinois. Four highly vigorous full-sibling triploid progeny were obtained, all of which had significantly lower leaf extension rates at low temperature relative to Mxg (3x) 'Illinois', and the two best siblings for leaf elongation also had significantly lower leaf CO 2 uptake rates and wholechain electron transport when grown at 10 °C/5 °C compared with Mxg (3x) 'Illinois'. Thus, hybrid vigour appears to have been insufficient to account for chilling tolerance in Mxg.
Doubling the chromosome number in triploid Mxg 'Illinois' resulted in hexaploid lines with significantly lower chilling tolerance with respect to A sat and Φ PSII (Figs 3-5; Supplementary Fig. S3 at JXB online). These results indicated that suboptimal ploidy can reduce chilling tolerance in Mxg even when advantageous genes and gene combinations are present. In previous studies, hexaploids of Mxg had slightly greater stem diameter, slower growth rates (Yu et al., 2009), and fewer shoots (Głowacka et al., 2010), compared with triploid Mxg. Thus, ploidy can affect a range of traits, including chilling tolerance.

What variation in chilling tolerance can be found in Mxg's parental species?
Based on the present sampling, Msa was typically a better source of chilling tolerance than Msi (Figs 1, 2; Supplementary  Fig. S1 at JXB online), though Msi from particularly cold environments in the northernmost part of the species natural range, such as northern China, and southeast Russia were not available for this study. The results were consistent with the fact that the natural range of Msa extends further north in Asia than Msi, indicating that Msa is generally better adapted to colder temperate environments than Msi (Clifton-Brown et al., 2008;Sacks et al., 2013).
Though only a few other studies of low temperature leaf elongation or photosynthesis have been conducted on multiple genotypes of Miscanthus, they all support the conclusion that there is genetic variation within Miscanthus for chilling tolerance (Earnshaw et al., 1990;Clifton-Brown and Jones, 1997;Farrell et al., 2006;Purdy et al., 2013). Farrell et al. (2006) compared four genotypes for leaf elongation of dormant rhizomes in the dark and found that Msi (2x) JESEL78 (a selection from the Jelitto population New Hybrid) was superior at all temperatures tested (7, 9, 11, 13, and 15 °C) to Mxg (3x), Msi (3x) 'Goliath' (GOFAL7), and Msa (4x) Sac-5 (='EMI-5'=MATEREC11). Purdy et al. (2013) studied leaf elongation at 12 °C for 4 d after transfer from 28 °C for four Miscanthus genotypes: Mxg (3x), Msi (3x) 'Goliath', Msa (4x) Sac-5, and Msi (2x) EMI-11. Purdy et al. (2013) observed no significant decrease in leaf elongation for Mxg, which may reflect the shorter duration of the chilling treatment than in the present study. All other genotypes in the Purdy et al. (2013) study had a significant decrease in leaf elongation relative to Mxg, which was similar to 32 (65%) genotypes tested in the present study. It is noteworthy that Msa (4x) EMI-5 was neither the best nor worst performing Msa under chilling conditions in the present study, but it was the only Msa included in the studies of Farrell et al. (2006) and Purdy et al. (2013). Clifton-Brown and Jones (1997) compared leaf elongation of Mxg with 21 accessions of Msi and four of Msa, and they observed that the best and worst genotypes under the lowest temperatures studied (5 °C for 11.5 h) were Msi. It is unclear why the study of Clifton-Brown and Jones (1997) did not identify Mxg as a superior genotype under the lowest temperature treatment, in contrast to the study of Purdy et al. (2013) and the present study. Earnshaw et al. (1990) observed that M. floridulus originating from high altitude (3280 m) in Papua New Guinea had greater chilling tolerance than a low altitude (2600 m) accession, as evidenced by greater chlorophyll fluorescence after chilling of leaf samples at 0 °C, which indicated that natural selection for adaptation has resulted in genetic diversity of chilling tolerance in Miscanthus. An important contribution of the present study is that it was possible to evaluate more than twice as many Msa genotypes as all of the previous studies combined, which enabled the observation that both diploid and tetraploid genotypes of Msa typically have high, though variable tolerance to chilling.
In the present study, some Msa had chilling tolerance equivalent to Mxg (3x) 'Illinois'. Tetraploid Msa 'PF30153', which was collected in Gifu Prefecture, did not differ significantly from Mxg (3x) 'Illinois' for chilling tolerance associated with leaf elongation and photosynthesis (Figs 1, 3, 4). Moreover, all Msa accessions studied, except the ornamental, chlorophyll mutant 'Gotemba Gold', did not show a significantly greater decline in the operating efficiency of photosystem II (Φ PSII ) relative to Mxg (3x) 'Illinois' at the end of the 11 d chilling treatment and upon recovery to warm temperature (day 12) (Fig. 3; Supplementary Figs S2, S3). Photosystem II is considered the most sensitive part of the photosynthetic apparatus to light-induced chilling damage and often the first manifestation of chilling damage (Long et al., 1994;Maxwell and Johnson, 2000). Thus, stability of the photosynthetic apparatus under cold stress is necessary but insufficient for high levels of photosynthesis per se at low temperature, yet the former trait may be of considerable value on its own. The ability to recover high rates of photosynthesis quickly upon return of warm temperatures may itself contribute to adaptation to fluctuating temperatures that are typical in temperate environments of the beginning and end of the growing season.
Given that Mxg (3x) 'Illinois' and most of the Msi included in the present study originated in southern Japan, yet chilling tolerance was highest for Mxg (3x) 'Illinois' and Msa, it is likely that Mxg's chilling tolerance was obtained from its Msa parent. Furthermore, as the tetraploid parent to this triploid, Msa has double the influence of Msi on the genetic makeup of the Mxg hybrid. Given that chilling tolerance varies among and within Miscanthus species, careful selection of parents for developing new Mxg bioenergy cultivars will be needed to ensure optimal adaptation to temperate environments.
High productivity in a cool temperate climate requires a genotype to be able to develop photosynthetically competent leaves under the chilling conditions of early spring. Without this, a crop canopy will be unable to use the high insolation of April, May, and June fully in the northern hemisphere. This capacity is fundamental to explaining the higher productivity of Mxg (3x) 'Illinois' relative to Z. mays (Dohleman and Long, 2009;Long and Spence, 2013). Based on a productivity model, Farrell et al. (2006) predicted that a Miscanthus accession with greater cold and chilling tolerance than Mxg could extend the growing season an average of 30 d, which would result in up to 25% higher yield. Thus, understanding how to select for improved rates of photosynthesis under low temperatures is key to improving productivity and sustainability of crops grown in temperate environments, such as Miscanthus. For breeding Mxg, choice of both the Msa and Msi parents will be important for achieving gains in chilling tolerance and concomitant yield potential in temperate environments. Indeed, the four new triploid Mxg progeny [Msa (4x) 'Bluemel Giganteus'×Msi (2x) var. condensatus 'Cabaret'] that were tested in the leaf elongation experiment had inferior chilling tolerance relative to their Msa parent, because their Msi parent was poorly adapted to cold temperatures.
What are the physiological mechanisms of chilling susceptibility and chilling tolerance of Miscanthus accessions?
All 16 accessions studied (13 Miscanthus, two sugarcane, and one maize) showed decreases in A sat , and Φ PSII , and increases in c i /c a , after transfer from 25 °C/20 °C to 10 °C/ 5 °C ( Fig. 3; Supplementary Fig. S2 at JXB online). Stomatal conductance for most of the genotypes (81% of accessions) increased on day 0 and then decreased during the following 11 d of chilling treatment. Since in the intercellular compartment the concentration of CO 2 increased as a consequence of a low rate of CO 2 fixation, the stomata reacted by reducing their aperture. Thus, differences among accessions in relative decreases in A sat cannot be explained by deficiency in CO 2 or by loss of stomatal function. Comparing the A sat reading with values of Φ PSII showed that light-induced chilling damage of photosystem II is the primary reason for low CO 2 assimilation in chillingsusceptible accessions. The most chilling-tolerant accessions retained relatively high Φ PSII and A sat during 11 d of chilling treatment and, upon return to warm temperatures (on day 12 of the experiment), they also recovered their gas exchange and chlorophyll fluorescence to pre-chilling treatment levels. These results are consistent with those of Farage et al. (2006), which indicated that the main mechanism behind avoiding the photodamage under chilling temperatures is the capacity for non-photochemical quenching of absorbed light energy. In contrast to the chilling-tolerant accessions, the chillingsusceptible accessions in the present study showed very low levels of recovery of measured parameters upon return to warm temperatures at the end of the experiment. This indicates a damaged photosynthetic system in susceptible accessions, and suggests the absence of alternative pathways for dissipating absorbed excitation energy under chilling conditions, leading to irreversible photo-oxidation of photosystem II (Allen and Ort, 2001).

Conclusion
Based on the sampling carried out here, Msa is a better source of chilling tolerance than Msi; thus an even more extensive survey of tetraploid Msa collected from diverse environments would be advantageous for identifying superior Msa parents and breeding new chilling-tolerant Mxg cultivars. There is also a need to identify Msi parents with superior chilling tolerance, though this may be more challenging than for Msa. To the extent that lessons from breeding Miscanthus for increased photosynthetic capacity at low temperature can be applied to other crops, especially other Saccharinae, great gains in agricultural productivity and sustainability may be obtained in temperate environments.

Supplementary data
Supplementary data are available at JXB online. Figure S1. Relationship between leaf elongation at warm and chilling temperatures for 51 Miscanthus accessions, two control sugarcane, and two maize lines. Figure S2. Leaf CO 2 uptake rate, quantum yield of photosystem II, stomatal conductance to water vapour, and ratio of intercellular to atmospheric CO 2 concentration for warm conditions prior to chilling treatment, after transfer of plants from warm to chilling (day 0), on the 11th day of chilling treatment, and 1 d after transfer of plants back to warm conditions. Numbers are expressed as a percentage of rates observed in warm conditions before the chilling treatment. Figure S3. Changes in quantum yield of photosystem II and intercellular to atmospheric CO 2 concentration following transfer of plants from warm to chilling conditions.