• Open Access

Thermal requirements for seed germination in Miscanthus compared with Switchgrass (Panicum virgatum), Reed canary grass (Phalaris arundinaceae), Maize (Zea mays) and perennial ryegrass (Lolium perenne)



    1. Bioenergy and Environmental Change, Institute of Biological, Environmental and Rural Sciences, Gogerddan, University of Aberystwyth, Aberystwyth SY23 3EB, UK
    Search for more papers by this author

    1. Bioenergy and Environmental Change, Institute of Biological, Environmental and Rural Sciences, Gogerddan, University of Aberystwyth, Aberystwyth SY23 3EB, UK
    Search for more papers by this author

    1. Bioenergy and Environmental Change, Institute of Biological, Environmental and Rural Sciences, Gogerddan, University of Aberystwyth, Aberystwyth SY23 3EB, UK
    Search for more papers by this author

    1. Institute of Biological and Environmental Sciences, School of Biological Sciences, University of Aberdeen, Cruickshank Building, St Machar Drive, Aberdeen AB24 3UU, UK
    Search for more papers by this author

    1. Bioenergy and Environmental Change, Institute of Biological, Environmental and Rural Sciences, Gogerddan, University of Aberystwyth, Aberystwyth SY23 3EB, UK
    Search for more papers by this author

    1. Bioenergy and Environmental Change, Institute of Biological, Environmental and Rural Sciences, Gogerddan, University of Aberystwyth, Aberystwyth SY23 3EB, UK
    Search for more papers by this author

Paul Robson, tel. +44 1970 823 091, fax +44 1970 820 241, e-mail: ppr@aber.ac.uk


The high establishment costs of Miscanthus by clonal propagation are a barrier to widespread deployment. Direct sowing is the cheapest method, but limited field trials have given generally poor results. Miscanthus, a perennial grass with C4 photosynthesis has tropical origins, but is found growing both at high latitudes (>40°) and altitudes (>1000 m) in Asia. In this paper, we investigate if significant variation in the thermal requirements for germination exist in 10 Miscanthus sinensis half-sib families and compare these with Panicum virgatum (Switchgrass – Trailblazer), Phalaris arundinaceae (Reed canary grass – P10) and Lolium perenne (perennial ryegrass cv AberDart) and maize (Zea mays cv Aviso). The comparisons were made on a thermal gradient bar with a controlled temperature oscillating ± 5 °C on a 12 h cycle and germination was monitored daily for 35 days at mean temperatures ranging from 5.3 to 26.5 °C. Base temperatures were calculated below which germination of at least 50% of viable seeds ceased. Base temperatures were lowest for L. perenne and Zea mays at 3.4 and 4.5 °C respectively; for different Miscanthus half-sib families base temperatures ranged between 9.7 and 11.6 °C and these were higher than maize and switchgrass which share C4 photosynthesis with Miscanthus. Parameters derived from germination and temperature were used to predict germination patterns in Europe based on historical climate data. We predict that seed establishment of Miscanthus in spring time is unlikely to be viable in Northern Europe under present climatic conditions without crop management practices aimed at raising soil temperature, and that useful variation in thermal requirement for germination in Miscanthus is available which should facilitate seed germination in other regions.


Mitigating climate change and the security of energy supply are the two main strategic drivers for the development of crops for conversion into energy and biofuels. Miscanthus spp. can produce high yields with low inputs of fertilizers and pesticides (Lewandowski & Schmidt, 2006). For Miscanthus, a key barrier is the lack of inexpensive and successful stand establishment (Christian et al., 2005; Lewandowski & Schmidt, 2006). To date most Miscanthus grown for biomass is a sterile triploid clone known as M. ×giganteus (Greef & Deuter, 1993). This clone produces proven high yields on good land and its sterility prevents weed risk arising from stray seed. However, clonal propagation, either through tissue culture or through rhizome splitting is expensive at around £1700 ha−1 (Jones, 2009) and is largely dependent on EU planting grants provided to some countries. We estimate the cost of sowing Miscanthus would be around £400 ha−1. Direct sowing is an inexpensive method of establishment widely used for grasses and cereals. Miscanthus species use C4 photosynthesis, are considered to be of tropical and subtropical origin, and therefore may have temperature requirements for germination that are not suited to seed propagation in more temperate environments. Trials in the UK with Miscanthus from a limited genetic base have shown that there is potential for seed propagation (Christian et al., 2005); however, in practice this method of establishment has proven unreliable (Naidu et al., 2003; Naidu & Long, 2004).

Miscanthus has been successfully grown in a wide range of European conditions and can be found at high altitudes and latitudes in Asia (Hsu, 1993; Clifton-Brown et al., 2008b). Light and fluctuating diurnal temperatures both enhanced germination (Hsu et al., 1985a; Hsu, 1989). Temperatures in the field near the soil surface undergo more extreme diurnal variations than air temperatures measured at 2 m in a screen (Geiger, 1950). It has been proposed that small seeded species detect these diurnal variations, and combined with light, these signals trigger germination (Thompson et al., 1977). To understand more about potential limitations on Miscanthus seed germination we evaluated the thermal requirements for germination of a wide range of Miscanthus accessions and to put these in context by comparison with other crop species that are direct seeded.

In the experiment described here, we simulated the thermal conditions of the seed bed by lowering the night time temperature by 10 °C, where the duration of day and night cycles was 12 h. We tested germination in a number of Miscanthus genotypes and compared them with germination in other candidate bioenergy grass species. A simple model was parameterized and, in conjunction with gridded meteorological data from the University of East Anglia Climate Research Unit (CRU) (New et al., 2002), was used to predict where seed propagation of Miscanthus is likely to be reliable compared with other grasses that could be used for energy production.

Materials and methods

Thermal gradient bar

An extensively used (Chorlton et al., 1997) thermal gradient system developed by Grant Instruments Ltd. (Cambridge, UK) was designed to test germination at temperatures along a gradient of 14 different temperatures with 14 accessions. In the original configuration, the set points of the hot and cold ends of the gradient bar were controlled by separate temperature controllers. In order to replicate more accurately field conditions during spring time in the United Kingdom, we modified this standard configuration to generate alternating 12 h day and night cycles with a diurnal shift of approximately 10 °C. The Grant controllers were replaced with a CR10 datalogger controller (Campbell Scientific, Loughborough, UK). A CR10 programme was written to control the hot and cold ends of the gradient bar, and nine calibrated thermocouples were distributed across the gradient plate on a grid to monitor system performance. Over the course of 35 days, the mean temperature at the cold end alternated between 0.1 (SD 0.35) °C during the cooling phase (night) and 10.1 (SD 0.19) °C during the warming phase (day) with a mean diurnal range of 9.9 (SD 0.26) °C. The hot end alternated between 21.6 (SD 0.48) °C at night and 31.9 (SD 0.37) °C in the day with a mean diurnal range of 10.1 (SD 0.33) °C. The temperature range between the hot and cold ends of the temperature bar was 21.6 °C. Mean cell temperatures were calculated by interpolating at cell centres using a second-order polynomial to describe the relationship between mean temperature recorded by the thermocouples and position on the gradient bar. The gradient bar was provided with supplemental overhead lighting from fluorescent tubes supplying 60–80 μ mol m−2 s−1 PAR which was provided on a 12 h cycle matching the day/night thermal cycle.

Plant material

In this test we included (1) ten half-sib families each derived from open-pollinated seed of a maternal M. sinensis genotype and a range of paternal plants from the UK Miscanthus germplasm collection, which comprises a wide range of Miscanthus genotypes of different morphologies (Clifton-Brown et al., 2008a) the M. sinensis maternal genotypes were selected on the basis of a high seed set phenotype of potential value to the breeding of seed propagated varieties (2) one commercial maize variety (Zea mays, cv Aviso) (3) one variety of switchgrass (Panicum virgatum Trailblazer, SWTB) (4) one genotype of reed canary grass (Phalaris arundinacea, variety #10 as produced for the EU reed canary grass project, P10) and (5) an Aberystwyth-bred perennial ryegrass (Lolium perenne cv AberDart), currently used in 40% of resown pastures in the UK) (Table 1). Seed harvested in autumn from open pollinations in the field, was stored after threshing at 4 °C for 6 months before being used in this study.

Table 1.   Details of germplasm tested on the thermal gradient bar, abbreviations used in figures and the text and mean seed mass (mg seed−1) determined on batches (n=14 unless indicated otherwise) of 30 seeds
IDSpeciesFurther informationMass (mg seed−1)
  • IBERS; Institute of Biological, Environmental and Rural Sciences, Miscanthus breeding programme.

  • *

    n=10 batches.

AvisoZea maysCommercial hybrid ‘Aviso’254.601.103
AberDartLolium perenneCommercial variety ‘AberDart’2.030.046
SWTBPanicum virgatumcommercial variety ‘Trailblazer’1.880.013
Mx36Miscanthus sinensisIBERS open pollinated cross1.440.021
Mx54Miscanthus sinensisIBERS open pollinated cross1.200.014
Mx128Miscanthus sinensisIBERS open pollinated cross1.170.017
Mx122Miscanthus sinensisIBERS open pollinated cross1.160.014
Mx129Miscanthus sinensisIBERS open pollinated cross1.080.014
Mx117Miscanthus sinensisIBERS open pollinated cross1.060.012
Mx116Miscanthus sinensisIBERS open pollinated cross1.010.019
Mx52Miscanthus sinensisIBERS open pollinated cross0.920.009
P10Phalaris arundinaceaPopulation 10 for EU project 1995-19980.910.008
Mx134Miscanthus sinensisIBERS open pollinated cross0.860.009
Mx24Miscanthus sinensisIBERS open pollinated cross0.63*0.019§

Seeds of each type were counted out into batches of thirty and their weight was determined. With the exception of AberDart, seed was surface sterilized in 1% hypochlorite for 1 min followed by at least 10 washes in an excess of sterile distilled water. AberDart seed had a husk that retained bleach, which in preliminary experiments inhibited germination, and therefore were not sterilized. Each seed type was randomly assigned to the 14 lanes on the thermal gradient bar and 30 seeds placed into each cell on the gradient bar. Owing to limited seed availability, Mx24 was only tested at the 10 highest temperatures. Water was supplied to the germinating seed from a capillary cloth made of industrial tissue paper.


On a daily basis for 35 days each seed was examined while it was gently rolled using tweezers. Germination was defined as 1 mm of radical emergence from the seed. Germinating seeds were removed from the test and recorded to calculate the number of seeds remaining in each of the 196 cells each day. After 35 days the number of viable seeds remaining was determined by raising the temperature set points at both the cold and warm ends of the gradient bar by 10 °C. The germination was recorded after a further 22 days. Adding this number to the number of seeds that germinated under treatment conditions gave the number of viable seeds.

Statistical analysis

For each cell on the gradient bar cumulative frequency distributions were fitted to accumulated daily counts of germinating seeds against time (days) using the CUMDISTRIBUTION procedure within GenStat® (Payne, 2010). Both logistic and Weibull distributions were evaluated and a lag phase was incorporated in each case. Since model fitting for neither distribution converged successfully for the greater majority of cells it was necessary to use a combination of both distributions. Where both models converged, choice of distribution was based on the lower residual mean deviance. Parameter estimates were used to estimate the time at which 50% of viable seeds had germinated (d50; days). Using the relationship between germination rate, the reciprocal of d50 (days−1), and mean cell temperature, the lowest mean temperature at which at least 50% of viable seeds could be expected to germinate (T0;  °C) was estimated by jackknifed linear extrapolation. The lowest mean temperature at which at least 50% of viable seeds could be expected to germinate within 15 days (T15d;  °C) was estimated in a similar manner.


For each 10′ grid of land mass within Europe monthly averages for daily minimum, mean and maximum temperatures and monthly precipitation and mean cloud cover were obtained from the Climatic Research Unit. Data are the result of interpolation between means of meteorological observations over the period 1961–1990 and include a correction for elevation within each 10′ grid (New et al., 2002). Monthly mean values assigned to the 15th day of each month were interpolated linearly to generate a daily time series for maximum, mean and minimum daily temperatures. Using the daily temperatures cumulative degree days relative to a threshold of 0 °C was calculated for each day of the year according to the method of Hastings et al. (2009b). Daily precipitation was calculated by dividing the monthly average by the rain days in the month and randomly distributing the rain days. Monthly Potential Evaporation (PET) for each 10′ grid was calculated from monthly mean temperature using the Thornthwaite & Mather (1957) method, and corrected for the aridity of the climate by the empirical correction for annual rainfall used by the Food and Agriculture Organisation of the United Nations (FAO) (Deichmann & Lars, 1991) to match the Penman-Monteith estimation. Actual PET (AET) was calculated by reducing the PET by a factor based upon the soil moisture content (Aslyng, 1965). The monthly PET values were divided by the days in the month to give daily PET values.

Soil field capacity and wilt point for each 10′ grid was extracted from the IGBP-FAO data set (FAO/IIASA/ISRIC/ISS-CAS/JRC, 2008) using ARCgis and plant available water was calculated as their difference. For each day the soil moisture content was calculated considering the soil to be at field capacity on the 1 January and adding or subtracting the daily difference between AET and precipitation. Any excess above field capacity was considered run-off.

In order to make good use of the growing season in the European Union we postulated that germination later than 15 May is too late for sufficient establishment before the next winter. Therefore, for a particular climate grid square, if mean temperatures for a 15-day period (matching T15d from the gradient bar) from the 1 January but before the 15 May are reached, the thermal requirements for germination are met. If soil moisture content was below the wilt point on this day it was considered that seed propagation would not be possible without irrigation. ARCGis was used to produce a map of the geographical area in Europe showing the limits of seed propagation for each of the grasses using these criteria.


Thermal gradient performance

An excerpt of 3 days out of the 35 days of the experiment of the measured temperatures at the hot, middle and cold points on the gradient bar is shown in Fig. 1. Mean cell temperatures over the course of the experiment ranged from 5.3 to 26.5 °C.

Figure 1.

 A 3 day sample of temperature cycles recorded on the thermal gradient bar (solid line). Dotted lines indicate the mean temperature recorded over the 35-day experimental period by thermocouples positioned at the midpoint and the warm and cold ends of the test area.

Seed size

Mean seed weight for M. sinensis varied between 0.6 and 1.4 mg (Table 1). Maize seeds were very large compared with the other grasses in this test with a mean weight of almost 250 times that of M. sinensis.

Germination profiles

Figure 2 shows percentage germination of viable seed from Miscanthus half-sib family (Mx117) exposed to average temperatures ranging from 5.3 °C to 26.5 °C (for clarity only data from 7 of the 14 temperature treatments are shown). With this particular half-sib family, at the lowest temperature no viable seed germinated and 100 per cent germination was only seen at the higher temperature treatments viz. mean temperatures of 19.4–26.5 °C.

Figure 2.

 Percentage germination of M. sinensis Mx117 as influenced by temperature (°C; denoted by symbols) and incubation time. For clarity only seven of the 14 temperatures applied are shown.

Viability and overall % germination vs. temperature

With the exception of two half-sib families (Mx129, 85.3% and Mx116, 88.4%) average percentage viability exceeded 92 per cent of available seeds (Fig. 3). There was no evidence of any systematic trend in viability over the temperature range examined. The percentage of viable seeds germinating after 35 days at different temperatures on the gradient bar was affected by temperature. At lower temperatures the numbers of seed germinating was reduced. Some Miscanthus sinensis seed germinated at the lowest temperature and all germinated at 7.2 °C but generally fewer than 10 per cent of viable seed. Percentage germination relative to viable seed was in excess of 90% for all seed at temperatures tested of 16.1 °C and above with the exception of two Miscanthus half-sib families Mx128 (80.8%) and Mx116 (83.3%). At temperatures tested of 9.1 °C and below, percentage germination of most seed lots was below 50% except for perennial ryegrass (86.7%) and switchgrass (55.2%). The highest precentage germination for Miscanthus at this temperature was 43.3% (Mx54).

Figure 3.

 Seed viability (% seeds; dotted line) and asymptotes of germination (% of viable seed; solid line) and mean temperature.

Lag time vs. temperature

The estimated lag time before visible germination increased with decreasing temperature treatment (Fig. 4). At higher temperatures lag times of 1–2 days were observed with the reference grasses and 1–3 days with Miscanthus. Some Miscanthus half-sib family reached lag times of below 3 days at lower mean temperatures, e.g. Mx54 14.4 °C, while others such as Mx129 required temperatures above 20 °C. Minimum mean temperatures, at which estimated lag times of <5 days were observed with maize, switchgrass, perennial ryegrass and reed canary grass, were 9.1, 10.9, 10.9 and 12.7 °C respectively; however, for Miscanthus seed mean temperatures of 14.5–19.4 °C were required. The lowest temperature treatment at which it was possible to estimate lag times for all seed tested was 16.1 °C. At 16.1 °C lag times of 1.2, 1.4, 1.8 and 2.7 days were calculated for reed canary grass, maize, perennial ryegrass and switchgrass respectively; with the exception of Mx52 all Miscanthus seed tested at this temperature had longer lag times.

Figure 4.

 Lag (days) before visible signs of germination and mean temperature (°C) as estimated from the fitted model.

Time to 50% germination

The number of days required for 50% of viable seeds to germinate (d50) decreased with increasing temperature treatment for all seed tested (Fig. 5). The lowest temperature at which d50 values could be calculated for all species, although not all Miscanthus half-sib families, was 10.9 °C, at this temperature d50 values for perennial ryegrass, maize, switchgrass and reed canary grass were 5.5, 7.8, 10.7 and 20.3 °C, respectively; d50 values for Miscanthus at this temperature ranged from 14.5 to 22.2 °C. There was no indication that any of the temperatures evaluated were supra-optimal with respect to germination rate for any of the seed assessed.

Figure 5.

 Days required for 50% of viable seeds to germinate and mean temperature (°C).

Estimates of base temperature (T0;  °C) for each seed type are shown in Fig. 6a. Base temperatures for the SWTB, P10, Aviso and AberDart were estimated at 8.0 (SE 0.65), 6.4 (SE 1.50), 4.5 (SE 1.34) and 3.4 (SE 0.61) °C, respectively. Estimates for Miscanthus half-sib families were higher [9.7 (Mx36) to 11.6 (Mx128 and Mx117) °C] than those for the reference grasses. However, these values represent the temperature at which germination rate is zero and without any constraint on the actual time to germination they are of limited relevance in a practical context. If we consider 15 days as an appropriate threshold time for 50% germination in the field, then practical estimates of minimum mean temperature requirements can be obtained by interpolating at a germination rate of 1/15 days−1 (Fig. 6b). These range from 12.0 (Mx54) to 16.9 (Mx128) °C for Miscanthus. Values for SWTB, P10, Aviso and AberDart were lower at 10.2 (SE 0.47), 11.2 (SE 0.98), 7.4 (SE 1.15) and 6.2 (SE 0.50) °C, respectively.

Figure 6.

 Estimated minimum mean (base) temperature required for at least 50% of viable seeds to (a) germinate (T0; °C) and (b) germinate within 15 days (T15d; °C). Vertical bars indicate the 95% confidence interval for each estimate.


Figure 7 shows the geographical limitation of rain fed germination and the northern limit of germination for each of the species considered. Miscanthus germination modelled using the genotypes in this study is predicted only to be successful in Southern Europe. Maize, a C4 species of subtropical origin has been subject to breeding programs for many decades and is able to germinate successfully as far north as Southern Sweden and Southern Finland. There is very little land in the 27 countries in Europe analysed in this study that is too dry for germination and even meditereanean and continental climates tend to have sufficent winter and spring rain to keep the soil moisture levels well above wilt point during the germination period up to 15 May.

Figure 7.

 Map of the southern limits of seed propagation due to limitations of rain fed moisture and the northern limit due to the mean temperature on 15 April. The orange areas represent regions that modelling demonstrated are suitable for seed propagation using the best performing Miscanthus half sib families plus all other species, the red areas are suitable for all Miscanthus tested plus all other species.


One of the attractions of switchgrass as an energy crop is the cheap establishment based on seeds (Khanna et al., 2008). Switchgrass and Miscanthus crops are barely removed from the wild especially when compared with maize and forage crops such as perennial ryegrass, which have been subject to intensive breeding effort (Bouton, 2008). There is one published report indicating that seed establishment in Miscanthus is possible but not without practical difficulties (Christian et al., 2005). The potential of achieving viable seed establishment for Miscanthus would be to produce a cheap reliably established energy crop that regularly produces higher yields than other grasses (Jain et al., 2010). The current study focuses on the potential of seed propagation under European conditions in terms of low temperature in spring, though this may not be the only limiting factor. We modified a static thermal gradient bar to simulate diurnal thermal conditions (Fig. 1). This was important because Matumura & Yukimura (1975) found that Miscanthus germination was higher when subjected to diurnal fluctuation in temperature and also a natural seed bed undergoes strong diurnal fluctuations in temperatures when the soil surface is freshly prepared for sowing. It was expected that germination in switchgrass would be lower than 50% because it is well known to have complex dormancy issues (Vogel, 2004). For this reason, switchgrass establishment protocols recommend that the seed is kept for 1 year at room temperature before sowing (Christensen & Koppenjan, 2010). Here, Trailblazer seed supplied by Ken Vogel of USDA, had high germination rates, generally >80% viable seeds, at temperatures ≥10.9 °C. Interestingly, Miscanthus sinensis seed appears to have almost no requirements for pretreatments to break dormancy, but if the seed is not chilled, germination start to decline after 6 months at room temperatures (Hsu, 1993). Under our experimental conditions, we found that all Miscanthus seed lots germinated well at average temperatures greater than 16.1 °C (typically >90%).

The base temperature (T0) where no germination occurred, derived from jackknifed estimates from the fitted model, was lowest in L. perenne (AberDart, 3.4 °C) (Fig. 6a). This matches literature reports for germination of L. perenne (1.4–2.9 °C, Moot et al. (2000); 3.5 °C Black et al. (2006); 3.7 °C, Larsen & Bibby (2005). Interestingly the estimated T0 value for germination of AVISO (a commercial Maize variety used in Europe) was lower than that for reed canary grass [Phalaris (P10)] which was selected for growth at higher latitudes (Olsson et al., 2004, Clifton-Brown et al., 2008a), an upland variety of switchgrass, had a threshold of 7.9 °C (Fig. 6a) which is lower than 10.3 °C reported by Hsu et al. (1985b) or 11.1 °C reported by Seepaul (2010). In our experiment T0 estimated for M. sinensis ranged from 9.6 to 11.6 °C, higher than all other seed lots tested. Our data concurs with reported values of T0 around 9–10 °C for M. sinensis (Hsu, 1985; Nishiwaki & Sugawara, 1997).

We found that temperature did not have a significant effect on percentage viability (Fig. 3). Temperature affected the speed of germination in all seed tested; decreasing temperature increased the time to germination. The time to germination at lower temperatures was longer in Miscanthus than switchgrass, reed canary grass, perennial ryegrass and maize. The longer lag period reflects the increased sensitivity of germination in Miscanthus seed to low temperatures. For field sowing, the minimum mean temperature base temperatures are more informative when combined with the lag time. For switchgrass recommended seeding rates range from 200 to 400 pure live seed m−2 in the southern and northern areas, respectively (Vogel, 1987; Christian et al., 2003). This represents over-sowing of up to 100 times the expected establishment rate, and assumes only a fraction of the sown seed become plants contributing to the final stand.

In this laboratory study we assumed practical germination limits of 50% germination within 15 days of sowing (T15d). T15d ranged from 6.2 °C for L. perenne to 17 °C for the most thermophilic M. sinensis Mx128. There was significant variation between the M. sinensis half-sib families with T15d ranging from 12 to 17 °C. The T15d estimates are not standard to seed germination protocols (e.g. AOSA, Association of Official Seed Analysts), but were here developed to provide a means to upscale the laboratory germination results to examine the spatial distribution of regions where these species could be expected to germinate before 15 May in any given year. The map in Fig. 7 shows where the different species tested are expected to have thermal conditions suitable for germination. If T15d in Miscanthus seed could be reduced to a value similar to that in reed canary grass, that is 0.7 °C lower, then the area where M. sinensis could be established in Europe is significantly increased (Fig. 7). If germination responses in Miscanthus approached that of AVISO maize, another C4 plant, then seed propagation in Miscanthus would be viable in most of Europe's current arable land including as far north as Southern Sweden and Southern Finland where maize is currently grown as a fodder crop.

Every modelling exercise requires assumptions. Here, we assume that temperatures required for germination are reached by 15 May. We reason that if germination does not occur by this date, then there will not be sufficient time remaining for seedlings to establish well enough to survive through the next winter [see also Vogel who makes a similar comment for switchgrass, (2004)]. At lower latitudes where the climate is warmer, it will be certainly possible to establish later than 15 May, but this does not affect our ability to show if germination is possible or not. A further assumption that we made relates to water availability, which is essential to germination. In this paper we have used a soil moisture deficit approach, which is the basis of many empirical plant growth models (Hastings et al., 2009b) to calculate when <75% of plant available water is left in soil at the time of germination. If this criterion is met, then the area is ‘too dry to germinate’ (purple shading on Fig. 7). The maps in Fig. 7 are indicative only and there is clearly a need to examine more carefully seed-to-soil hydraulic contact which will depend on many factors including soil textures, seed bed preparation (tilth), and planting depth (Christensen & Koppenjan, 2010). The depth of sowing will also impact on the temperature of the seed, here we used air temperatures to model germination but topsoil temperatures of bare earth have been shown to exhibit larger diurnal temperature variation in spring conditions (Geiger, 1950). Germination will also be affected by the pH of the topsoil (Aso, 1976), this was not a feature of this experiment.

The need for seed based establishment systems for perennial bioenergy crops is urgent in order to speed up the deployment of bioenergy crops and reduce the cost of crop establishment in terms of both financial and energy costs. In Europe, targets have been set to produce 80% of the RTFO (Renewable Transport Fuel Obligation) by 2050. In the United States, George Bush included the development of second-generation energy crops in his 2006 State of the Union address (Bouton, 2008). This included a commitment for extra funding for cutting edge research on crops such as switchgrass. One of the limitations of growing Miscanthus is the large cost of crop establishment by propagation, either by micropropagation or by production of rhizomes. In each case the financial, energy and greenhouse gas cost reduces the sustainability of Miscanthus as an energy crop (Clifton-Brown et al., 2007; Hastings et al., 2009a; Jain et al., 2010). In addition, propagation by seed is more readily scalable to rapidly cope with the potential increases in area of biomass production particularly from new varieties with improved yield or quality characteristics. Currently the UK and other EU countries provides grants for the establishment of perennial energy crops such as Miscanthus and short rotation coppice willow or poplar to reduce the farmers up-front costs and cover the loss of income for the establishment years. This is not sustainable in the long term although appropriate to establish a new industry. As carbon trading becomes more mature and carbon credits increase in value such incentives may no longer be necessary. This process may be hastened by legislation to reduce GHG emissions by 80% by 2050 in the United Kingdom and energy security legislation such as the USA Energy Policy Act (EPAct) 2005.

The data presented in this paper have focussed on Miscanthus in Europe; however, the model developed may be applicable to other regions adopting biofuel crop production. The data presented show that thermal requirements of germination currently limit crop establishment via seed to more southerly latitudes. Further laboratory experimentation is needed to determine if there is more variation in thermal responses to temperature than we have found here by including M. sacchariflorus and further M. sinensis accessions from a wide latitudinal and longtitudinal gradient. If low thermal requirements were found these could be exploited in breeding Miscanthus suitable for seed-based establishment in much cooler regions such as Northern Europe. This is a long-term objective. A shorter term objective, which could respond quickly to the urgent policy needs, would be to make changes to crop agronomy. Crop management practices such as the use of films or fleeces or organic top dressing can increase the mean temperature of the soil during after sowing and encapsulating the seed in dark organic material can achieve the same, while supplying an optimum germination environment. In the past 10 years, cheap biodegradable films have been used to accelerate maize germination and establishment (Keane et al., 2003). Plastic films allow earlier sowing, result in faster establishment rate and retain moisture and have been so successful, that commercial machinery is widely available that can lay plastic film at low cost and high speed (e.g. Samco in Ireland). The application of agronomic changes such as plastic films along with genetic improvement in seed germination will undoubtedly have a significant impact on the establishment of Miscanthus crops. Seed production is not addressed in this paper. Seed set is quite high in Miscanthus: Nishiwaki & Sugawara (1996) reported estimates of seed rates of 6500–8000 seeds per m−2 and in Asia, mean number of seeds per inflorescence were reported at over 1000 (Hayashi, 1979), ideally any improvements in thermal requirements for germination would not significantly impact on seed set.

This paper demonstrates the poor germination rates in M. sinensis at lower temperatures and explains why seed trials of M. sinensis have often resulted in low levels of plant establishment. The modelling work provides an overview of the relative performance of different crops based on the thermal responses of germination and demonstrates the potential spatial impact of small changes in the thermal requirements for germination. The potential genetic variation has been discussed, and a pragmatic interim solution has been proposed using biodegradable films. The application of agronomic developments, like plastic films, and genetic improvements through analysis of germination phenotypes such as those reported here suggest the efficiencies available from seed-based establishment may be attainable in cooler regions even for thermophilic crops. Further work is needed on the effect of soil temperature under field conditions and on the temperature raising effects of different plastic films, as well as other factors such as seed bed characteristics and depth of sowing before guidelines can be given to farmers and agronomists.


The authors would like to thank Athole Marshall for permission to modify his gradient bar, Charlotte Hayes for provision of the seed and Mark Hirst for his help with the experimental design and preliminary data analysis. This work was funded by DEFRA contract NFO426 and by Biotechnology and Biological Sciences Research Council (BBSRC) Institute Strategic Programme Grant BBSEG00003134.