• Open Access

Future energy potential of Miscanthus in Europe



    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. Biomass and Biorenewables Programme, Institute of Biological, Environmental and Rural Sciences, Aberystwyth University, Plas Gogerddan, 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. College of Physical Sciences, University of Aberdeen, Fraser Noble Building, Kings College, Aberdeen AB24 3UE, UK,
    Search for more papers by this author

    1. Institute of Ecology, University of Innsbruck, Technikerstrasse 25, 6020 Innsbruck, Austria
    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

Astley Hastings, tel. +44 1224 272 702, fax +44 1224 272 703, e-mail: astley.hastings@abdn.ac.uk


European field experiments have demonstrated Miscanthus can produce some of the highest energy yields per hectare of all potential energy crops. Previous modelling studies using MISCANMOD have calculated the potential energy yield for the EU27 from mean historical climate data (1960–1990). In this paper, we have built on the previous studies by further developing a new Miscanthus crop growth model MISCANFOR in order to analyse (i) interannual variation in yields for past and future climates, (ii) genotype-specific parameters on yield in Europe. Under recent climatic conditions (1960–1990) we show that 10% of arable land could produce 1709 PJ and mitigate 30 Tg of carbon dioxide-carbon (CO2-C) equivalent greenhouse gasses (GHGs) compared with EU27 primary energy consumption of 65 598 PJ, emitting 1048 Tg of CO2-C equivalent GHGs in 2005. If we continue to use the clone Miscanthus×giganteus, MISCANFOR shows that, as climate change reduces in-season water availability, energy production and carbon mitigation could fall 80% by 2080 for the Intergovernmental Panel on Climate Change A2 scenario. However, because Miscanthus is found in a huge range of climates in Asia, we propose that new hybrids will incorporate genes conferring superior drought and frost tolerance. Using parameters from characterized germplasm, we calculate energy production could increase from present levels by 88% (to 2360 PJ) and mitigate 42 Tg of CO2-C equivalent using 10% arable land for the 2080 mid-range A2 scenario. This is equivalent to 3.6% of 2005 EU27 primary energy consumption and 4.0% of total CO2 equivalent C GHG emissions.


Miscanthus is a rhizomatous C4 grass, which is a native of SE Asia and the Pacific Islands. The naturally occurring interspecific hybrid Miscanthus×giganteus (Hodkinson & Renvoise, 2001) was identified as a potential high-yielding energy crop in Europe (Sloth, 1985; Nielsen, 1987; Greef & Deuter, 1993; van der Werf et al., 1993). Field trials have shown that for many locations in Europe M.×giganteus has the largest energy yield of all potential bioenergy crops in terms of net MJ ha −1, and the highest energy-use efficiency (EUE), in terms of the energy cost of production, due to its relatively high yields and low inputs (Sims et al., 2006; Heaton et al., 2008).

Stampfl et al. (2007) used the MISCANMOD model, developed by Clifton-Brown et al. (2000, 2004), and mean meteorological parameters for the period 1960–1990 to show that M.×giganteus has the potential to provide a high proportion (39%) of the 2020 European targets for renewable energy.

Hastings et al. (2008) rewrote MISCANMOD in fortran (renaming the model MISCANFOR) to overcome some of the limitations associated with using ms excel®. Later MISCANFOR was modified to incorporate improved process descriptions for the environmental impacts on radiation-use efficiency (RUE) of soil capillary pressure water stress, ambient temperature and leaf formation temperature (Hastings et al., 2009a).

Under current and future climates, drought and frost resistance are the key traits needed to extend the range of Miscanthus in Europe. In Asia, Miscanthus grows in a diverse range of climates from the near Arctic to the tropics, and we expect this will lead to new superior hybrids (Clifton-Brown et al., 2008).

In this paper, we calculate the greenhouse gas (GHG) emissions, energy consumption and energy yield from production and use of Miscanthus for bioenergy for four future climate scenarios for EU27 countries.

Previous predictions of energy yields and carbon mitigation have been made using the mean meteorological data of the period 1960–1990 (Clifton-Brown et al., 2004; Stampfl et al., 2007) or have used the mean-predicted conditions for future scenarios (Hastings et al., 2008). This approach does not estimate how the interannual variability in temperature or rainfall will affect the sustainability of the yield performance over the crop life (here estimated from M.×giganteus to be 15 years). It is therefore important when making crop viability and yield predictions for future climate scenarios to generate representative meteorological time series that reflect likely annual climate variations and to identify the probability of climatic events which could kill the crop such as prolonged drought or deep soil freezing.

In this paper, we describe a study using MISCANFOR to explore possible energy production from M.×giganteus and a new hi-tech hybrid based on traits derived from a well-characterized Miscanthus sinensis hybrid. Changes in the potential geographical range and the annual contribution of energy in the EU27 of Miscanthus for biomass were investigated using future climate scenarios. By modelling the growth of this hi-tech hybrid, we show how by breeding in advantageous traits that have been observed in different existing genotypes, the range of environments that Miscanthus could be commercially grown changes when compared with those possible for M.×giganteus.

Materials and methods


The Miscanthus growth model MISCANFOR. MISCANMOD (Clifton-Brown et al., 2004) was rewritten in fortran and further developments were made by improving the conditioning of input meteorological variables and the soil water deficit calculation. A water stress function based upon soil water capillary pressure, a temperature-dependant RUE function and a plant physiological stage model using water deficit, frost-induced senescence and photoperiod were included. The new model is called MISCANFOR and is described in detail in Hastings et al. (2009a).

The GHG emission model. A GHG emission and energy cost model for growing and using Miscanthus as an energy crop was based on St Clair et al. (2008) for the cropping cycle, and Hastings et al. (2008) and Clifton-Brown et al. (2007) for the soil GHG emissions. The cropping cycle costs in both energy and GHG emission terms are assessed for three stages: crop establishment, annual crop maintenance and yield-related costs.

The crop establishment energy and GHG emission cost includes ground preparation, machinery and initial rhizome production. This cost is apportioned over the life of the Miscanthus crop which is considered to be 15 years (Clifton-Brown et al., 2007). Annual crop cycle costs are divided into two parts: one that is land area specific, such as herbicide and fertilizer applications, and those that are yield specific, such as fertilizer quantities, harvesting, transportation, storage and drying costs (Lai, 2004; Lewandowski et al., 2003; Clifton-Brown et al., 2007; St Clair et al., 2008).

In addition to the anthropogenic GHG emissions associated with the production of Miscanthus, we also consider the soil GHG emissions, which depend on the initial soil organic carbon (SOC) at the time of crop establishment and the estimated SOC at the end of the crop cycle. The GHG emissions of CO2 are calculated as the SOC gain or loss prorated over the 15-year crop cycle (Jørgensen et al., 1997; Kahle et al., 2001; Lewandowski & Schmidt, 2006; Hastings et al., 2008). N2O emissions are calculated using the default tier 1 Intergovernmental Panel on Climate Change (IPCC) emission factor for N2O emissions from N fertilizer application (IPCC, 2007), plus a function of any decrease in the SOC over the 15-year crop cycle (Hastings et al., 2008).

The CO2 and N2O emission model is based upon a study that parameterized the Denitrification Decomposition (DNDC) model (Li, 1995) for Miscanthus, based upon published crop experimental data, where the SOC was measured during the course of a Miscanthus crop cycle (Beuch et al., 2000; Kahle et al., 2001; Hansen et al., 2004; Heaton et al., 2004; Clifton-Brown et al., 2007). Using a sequence of hypothetical initial SOC values and crop yields, this model was then used predict the cumulative GHG emissions over the cropping cycle and the SOC at the end of the cycle. From this data, a four pool soil organic material (SOM) model with exponential decay time constants of 0.1, 2, 30 and 500 years (Paustian et al., 1992; Brady & Weil, 2002) was constructed in excel using the root and harvest debris as the organic material input (cohort) for the year. This four pool model is used to calculate the decay of the organic input for each year, which is considered as a separate cohort to the initial SOC following the theorem of Bosatta & Ågren (1991). This results in 16 cohorts for the last year of the Miscanthus crop cycle. The model calculates the SOC at the end of each year and the net emission or mitigation of carbon through soil respiration. The initial SOM does not contain the most labile pool but the litter does. Summing the emissions or sequestration of carbon for the 15-year crop cycle, an expression relating the annual SOC reduction or accumulation to the initial SOC level and average annual dry matter (DM) yield was developed. Similarly, an expression relating the N2O emissions in the case of a net SOC reduction was derived from the DNDC simulation. N2O emissions relating to fertilizer use are discussed later.

Energy intensity model. The EUE is calculated as the inverse of the ratio of the energy in MJ (of diesel, fertilizer, herbicides, pesticides, crop management and transport) consumed during growing and processing Miscanthus fuel to produce 1 MJ energy in the furnace, and this is compared on the same basis with those of bituminous coal, diesel oil and natural gas (MJ MJ−1) for each year and grid block. Crop energy intensity is the amount of useful energy that can be produced on 1 ha of land. Useful energy is the actual energy minus the energy used for growing and processing the fuel. For each grid point/year, the carbon intensity (CI), which is the total GHG emissions expressed in carbon dioxide-carbon (CO2-C) equivalents using the established global warming potentials for CO2, CH4 and N2O of 1, 23 and 296 over 100 years (IPCC, 2007) are calculated as kg CO2-C eq. MJ−1. As with EUE, GHG emissions from Miscanthus are then compared with the fossil sources coal, oil and gas (Sims et al., 2006).

Model verification of energy efficiency and GHG emissions. To estimate the carbon cost of growing, storing and transporting a Miscanthus crop, we applied the methodology of Lai (2004) using information from St Clair et al. (2008), Lewandowski et al. (2003), Clifton-Brown et al. (2007) and the UK Department of the Environment, Food and Rural Affairs (DEFRA) recommended crop management system for Miscanthus, to construct two scenarios for crop production and use. The ‘optimum’ scenario considered local use of the crop within 20 km, harvested nutrient replacement with fertilizer application, two pre-emergence applications of herbicide (in year 1 and 2) and using rhizome plant propagation. This resulted in a fixed energy cost of 5.64 GJ ha−1 yr−1 and a yield-related cost of 0.61 GJMg−1 yr−1. We also consider Miscanthus fuel to have a moisture content of 30% (this is the lowest moisture content that can be achieved on field) that reduces the energy yield by the latent heat of vapourization of water of 2.72 MJ kg−1 during burning or drying.

The ‘current practice’ scenario used in this analysis relates to current application and practices for Miscanthus biomass fuel, for which data is currently available. This uses micropropagation of plants, two herbicide applications per crop life, nutrient replacement fertilization and the collection of biomass for a large remote cofired coal power station with a mean transportation distance of 300 km. In this case, the energy cost increased to a fixed cost of 9.1 GJ ha−1 yr−1 and a yield-related cost of 2.3 GJ Mg−1 yr−1. We used the ‘current practice’ criteria in this study to calculate the net energy yield for all the future climate scenarios, along with the assumption that this moisture level would be the minimum achievable by natural field drying. Assuming that in the future the ‘optimum scenario’ is achievable in this study we model the benchmark case against which future improvements can be measured.

Using the energy-use scenarios, we calculated the GHG carbon equivalent emissions cost as 589 kgC ha−1 yr−1 and 195 kg C Mg−1 yr−1 yield, using the same ‘current practice’ scenario and 118 kg C ha−1 yr−1 and 65 kg C Mg−1 yield yr−1 for the ‘optimum scenario’. These included emissions from the use and manufacture of machinery, herbicides and fertilizers.

A simplified relationship for soil carbon emissions that related annual CO2-C equivalent emissions (CEE) in Mg C ha−1 yr−1 to annual dry matter yield (DMY) in Mg ha−1 yr−1 and the initial SOC in Mg C ha−1 was derived by use of the DNDC model (Li, 1995) which was shown to match SOC values well over Miscanthus 15-year crop cycles and typical grasslands (Foereid et al., 2004; Hansen et al., 2004; Clifton-Brown et al., 2007):


Using the same methodology the N2O-N emissions caused by the reduction in SOC (CNEE) are found to be related to the CO2-C emission emanating from this change by the following relationship if CO2 emissions (kg C ha−1) is a positive value:


Nitrogen chemical fertilizer is applied at the rate to replace the nitrogen removed by the harvest. This is calculated as 0.3% of the weight of harvested Miscanthus dry matter (Lewandowski et al., 2003). The N2O emissions associated with fertilizer use are calculated using the default tier 1 IPCC emission factor for N2O emissions from N fertilizer application (IPCC, 2007). The climate forcing related to CO2 used for N2O is 296 (IPCC, 2007).

These relationships were used in the calculation of the carbon cost of the Miscanthus crop for each grid block. From the net energy yield and the carbon emissions the CI of the resulting Miscanthus fuel is determined as kg C MJ−1.

Sustainability flags. At each grid point and scenario time slice, if the CI of the Miscanthus energy crop at that grid point is greater than that of bituminous coal (0.033 kg C MJ−1) used in the furnace, we consider that it is not sustainable to use that grid point for Miscanthus production and a ‘not sustainable’ flag is generated. Similarly, if the energy required for growing and harvesting the crop is more than the energy generated at the furnace then the ‘not sustainable’ flag is also generated. We also calculate the EUE (EUE=useful furnace energy output/energy cost of production) and CI in kg CO2 eq. C MJ−1. An EUE of 1 is equivalent to 4.5 Mg ha−1 harvest yield for most soil and climatic conditions. Because Miscanthus cropping requires land resources, if grown on arable land it will be in competition with food and other nonfood crops. Currently European food production is in excess requirements (Eurostat Data Base, 2008) so an economic criteria for sustainability would be 7 Mg ha−1 (Stampfl et al., 2007) to include a profit element for the farmer. We have not attempted to model a balance between the land demands for world food production and bioenergy in this paper.

Data sets

The geographic window. The geographic window, within which the study was performed, ranged from longitude 11°E to 41°W and latitude 33 to 71°N. This area covers all European countries up to the Russian border (Cyprus is in the EU27 but not geographically in Europe). All model runs were made using data prepared on a 5′× 5′ grid over this area.

Climate data 1900–2000. Monthly precipitation (Mpt), average daily maximum temperature (Maxt), average daily minimum temperature (Mint), average daily mean temperature (Meant), average daily temperature range (TmpR), average daily cloud cover (Mcc), frost days (Mfd), precipitation days (Mpd) and average daily vapour pressure (Mvp) values for each 5′× 5′ grid cell were extracted from the 0.5° global climate data set provided by the Climate Research Unit, University of East Anglia. These monthly values are based upon spatially interpolated observed data from a network of weather stations (Mitchell et al., 2004).

Climate data 2000–2100. For future climate scenarios, Mpt, Meant, TmpR, Mcc and Mvp for each 5′× 5′ cell were extracted from the 0.5° global climate data sets provided by the Climate Research Unit, University of East Anglia (Mitchell et al., 2004). Monthly values were provided based upon outputs from the HadCM3 global climate model forced by four IPCC CO2 emissions scenarios, as reported in the special report on emissions scenarios (SRES) (Nakićenovićet al., 2000). The four climate scenarios examined (based upon the emissions scenarios) were A1FI (world markets-fossil fuel intensive), A2 (provincial enterprise), B1 (global sustainability) and B2 (local stewardship).

To analyse the interannual variability of yields and the probability of drought and frost kill events for future scenario time slices at 2020, 2050 and 2080, a representative 102-year time series of monthly meteorological data was used to run the model. This is constructed by adding the difference between each monthly meteorological variable for the scenario year and the mean value for the 1960–1990 time period to each month of the Climate Research Unit (CRU) 1900–2002 meteorological time series. This process was repeated for each grid block for the all the meteorological parameters used (Meant, TmpR, Mpt, Mvp and Mcc).

Potential evaporation (PET) and photosynthetically active radiation. Monthly PET for each cell was calculated from monthly mean temperature (Meant) 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 Organization (FAO) (Deichmann & Lars, 1991) to match the Penman–Monteith estimation.

Monthly global radiation (GlobRad) for each cell was calculated using latitude, day of year, Mvp and Mcc using the SWAT2000 method (Arnold & Foher, 2005). This method includes corrections for solar distance, daily declination and latitude. photosythetically active radiation (PAR), incident radiation between 400 and 700 nm or incident radiation absorbed by the plant, is then calculated from GlobRad using vapour pressure deficit to determine the partitioning of energy, and using the leaf area index calculated the previous day (from MISCANFOR) to calculate the albedo and to determine reflected energy.

Daily meteorological input variables for MISCANFOR. For each grid point and year, the monthly time series for Mpt, Maxt, Mint, PET and PAR were used to generate a daily time series for the year to be used as an input explanatory variable for MISCANFOR. Maxt and Mint were linearly interpolated between the monthly averages to provide daily values (Muselli et al., 1999; Hastings et al., 2009b). Mfd data were used as input variables for the historical years to calculate frost kill events and the growing season for the historical data, but were not available for the scenario data. For Mpt and Pet the average daily rainfall was calculated based on the number of Mpd for historical data and the number of calendar days for the scenario data for which Mpd was not available.

Soil data. Global gridded surfaces of selected soil characteristics (IGBP-DIS) were available from the International Geosphere–Biosphere Programme – Data and Information System Data set [available on-line (http://www.daac.ornl.gov) from the Oak Ridge National Laboratory Distributed Active Archive Centre, Oak Ridge, TN, USA]. Wilt point (Wp), field capacity (Fc), bulk density (Rhob), soil organic carbon to 1 m (SOC), thermal capacity (Tc) and total nitrogen (Tn) were available on a 5′× 5′ grid covering the whole world. This grid was used as the model reference grid.

Land use data. Land cover data for European countries except Norway, Switzerland, Belarus, Ukraine Moldavia and Serbia are available in a 250 m × 250 m grid in the CORINE Land Cover 2000 (CLC, 2000) data set (EEA, 2002). The land use is divided into 44 categories of which only four are arable cropland suitable for Miscanthus cultivation (e.g. Miscanthus cannot be grown as an undercrop in orchards). arcgis was used to calculate the proportion of each 5′× 5′ grid block which was arable cropland by assigning each (CLC, 2000) cell a value of 100 if it represents an arable category and 0 if not and performing a bilinear interpolation of the values to the 5′× 5′ grid to obtain the percentage of arable land in that grid. This value was then converted to hectare of arable land per grid block for the (CLC, 2000) baseline land use data set.

Future land use scenarios were constructed according to Rounsevell et al. (2006) for 2020, 2050 and 2080 which also used CORINE baseline land cover. Compared with the baseline data for arable land, the future scenarios include land used for liquid, nonwoody and woody bioenergy crops, as well as surplus land with no socio-economic purpose. The scenarios were available for EU15 plus Norway and Switzerland. Baseline totals for arable land were available for all EU27 countries. Here, we consider that all the land available for energy crops is used for Miscanthus cropping to establish the upper limit for possible energy production from Miscanthus. These scenarios establish the area of land available for energy crops in each country for each scenario and each time slice. For each country and time slice, the ratio of the area under energy crops compared with the baseline arable land (BFrat) in that country is calculated. In order to calculate the energy yield for each country, at each time slice, we consider that the distribution of Miscanthus crops are the same as current arable land, and we prorate the total country yield possible using all arable land by the percentage of arable land available for bioenergy in that country per scenario/time slice predicted by Rounsevell et al. (2006). For newly joined countries to EU27, we consider that that each country will have available a percentage of arable land that is the mean arable land percentage for EU15 (Rounsevell et al. 2006).

Growing range limitations. As mentioned above the range where Miscanthus can grow successfully is limited by frost and drought that can kill the plant. For each grid point, we calculate the number of plant kill events that occur during the time series being investigated. Here, we assume drought kill occurs after the soil moisture content falls below the wilt point for >60 days for M.×giganteus, derived from field observations by Schwarz and Greef of M.×giganteus growing on a rain-fed sandy soil in central Germany and unpublished pot experiments by Clifton-Brown. Some M. sinensis varieties under water-limited conditions regulate transpiration through stomatal closure and stay green and then resume growth as soon as soil water becomes available again (Clifton-Brown & Lewandowski, 2000b; Clifton-Brown et al., 2002). Here, we assume the new hi-tech hybrid can tolerate up to 120 days below the wilt point.

Winter frost kill events occur when soil temperatures fall to levels that are lethal to the overwintering rhizome. Soil temperatures were modelled using Fourier's law of heat conduction, the FAO-IGBP soil properties of Rhob and Tc, the soil moisture from MISCANFOR, a geothermal gradient of 0.01°K m−1, a heat flow of 3 × 10−2 W m−2 and using the explanatory variables of Meant, TempR and Mfd. This showed that 30 days with a mean temperature <−3.4 °C causes the soil temperature to fall <−3.4 °C, the threshold for M.×giganteus kill (Clifton-Brown & Lewandowski, 2000a), and 30 days with a mean temperature <−6 °C (Clifton-Brown & Lewandowski, 2000a), for M. sinensis (here incorporated into the new hi-tech hybrid) kill. Simulations showed that this did not vary much with soil type but did vary with water content, but as in most climates the soil is at field capacity during the winter frost we did not consider this in the model. These criteria were used to calculate the frost kill flags for each year. As current estimates of the life of a M.×giganteus crop is 15 years, a criteria of one kill events per 15 years is used to limit the range of both Miscanthus genotype.

For the base case, the period 1960–1990 was used to calculate the kill events and for future time series the entire 102-year, generated time series was used. For the purposes of yield calculations, if a kill event occurs then the yield of the following year is set to zero to model the re-establishment of the crop.

Technology change data. Miscanthus development in terms of breeding, agronomy and energy utilization, is still in its infancy (Clifton-Brown et al., 2007). To date, most information is available for the sterile hybrid M.×giganteus from trials in Europe, which is a wild accession from Japan. For this study, we use parameters derived mainly from M.×giganteus plot trials. A small European trial network with five sites on a wide latitudinal spread and 15 genotypes has provided some additional parameters for drought and frost tolerance in a M. sinensis hybrid (Clifton-Brown et al., 2001). This hybrid was 21% lower yielding than M.×giganteus in southern Germany where climatic conditions were very favourable without drought. Here, we make the conservative assumption that with breeding, it will be possible to combine the improved drought and frost tolerance traits in a hybrid that yields as well as M.×giganteus on all sites. In this paper, we refer to this as the ‘theoretical new technology hybrid’ (hi-tech hybrid).

Running the model

Model framework. The MISCANFOR plant growth module is encapsulated in a model (Fig. 1) that outputs the predicted variables of Miscanthus peak aboveground dry matter (PDM) and harvest dry matter (HDM) yield in Mg ha−1, the net (NEY) and gross (GEY) energy yield GJ ha−1 and the GHG emissions (GHGE) in CO2-equivalent carbon in Mg C ha−1 on a 5′× 5′ resolution grid over the geographical area of Europe.

Figure 1.

 Model framework used to predict energy production and greenhouse gas (GHG) mitigated by growing Miscanthus in Europe as an energy crop. The model uses MISCANFOR (Hastings et al., 2008) to predict Miscanthus dry matter harvest yields and an exponential decay model (Hastings et al., 2008) to predict GHG emissions. (1) Climate Research Unit (CRU) (Mitchell et al., 2004). (2) Food and Agriculture Organization (FAO) (Global Soil Data Task Group, 2004). (3) CORINE land use cover data set (CLC, 2000). (4) Rounsevell et al. (2006). (5) Greenhouse gas. (6) arcgis geographical mapping software from Environmental Systems Research Institute Inc.

Simulation procedure. The complete grid model, using the CRU meteorological data for the appropriate year and the IGBP soil data for the appropriate grid reference the model output was compared with the experimental plot data described by Hastings et al. (2009a) and a linear regression used to compare the model with the experimental data.

The base case was constructed from the simulation run on the period 1960–1990. The predicted variables are averaged over this period and displayed as the base case using the figure of 10% of arable land, corresponding to the historic set aside area in the EU, as a theoretical indicator to calculate potential yields and mitigation for current conditions. From the time series of predicted variables on each grid block over the time period, the mean, 95% confidence interval and standard deviation (SD) of PDM, HDM, NEY, GEY and GHGE were calculated to predict the variation in the contribution of Miscanthus to European energy supply and carbon mitigation that could be expected due to the variation in annual meteorological conditions.

For future climate scenarios, predicted variables were calculated for time slices 2020, 2050 and 2080 for each of the four climate change scenarios. As future climate scenarios are trends with no interannual variation, the model was run on the 102-year time series generated for each scenario-year time slice. In this way the interannual variation of the predicted variables could be estimated and the mean and SD of the yields and mitigation calculated. The meteorological conditions which result in kill events could be also calculated for this time series, to investigate the crop sustainability and growing range.

Presentation of results. The predicted values of each variable for each grid cell for each time slice and scenario were multiplied by the arable land area grids and scenario percentage of land available for bioenergy, to calculate the total energy yield and GHG mitigation for each grid block. The grid values were then summed for each scenario to produce energy yields, GHG mitigation and CI of Miscanthus per country.


Model verification of peak yield predictions

Hastings et al. (2009a) have previously shown that the MISCANFOR Miscanthus growth model predicts the results of the European plot and field experiments with a linear unity slope (r2=0.84, n=36). Here, we run the model using the CRU meteorological data and IGBP-FAO soil data on a 5′× 5′ grid and repeat the results for the year and geographical location to prove that the model system achieves similar peak autumn yield predictions P<0.001. To achieve this match in Mediterranean climates we had to add a photoperiod function that started crop growth after day length exceeded 12 h.

The base case scenario using historical data 1960–1990

The base case 1960–1990 time series was run using the M.×giganteus parameters and the mean peak DM yield for the 31 years was calculated and is displayed in Fig. 2a. This yield was used to calculate the net energy yield, which considers 67% of the peak yield converted to gross energy using an energy intensity of 18 GJ Mg−1 (Clifton-Brown et al., 2007). This is reduced by the energy cost of production and the latent heat of the moisture to calculate the net energy and EUE. If the EUE is less than unity, the grid block was excluded from country energy and C mitigation calculations; the excluded blocks are shown in Fig. 2c in light blue. The CI of the energy crop is also calculated and grid blocks with a CI worse than coal are displayed in Fig. 2c in black.

Figure 2.

  Miscanthus×giganteus mean peak dry matter yields for the period 1960–1990 calculated using MISCANFOR (a) compared with the distribution of Arable land in 2000 for EU27 (d). (b) Dry-matter yield map with frost kill mask (blue) and drought kill mask (brown) superimposed. The CORINE (2005) land use data set on a 250 m × 250 m grid is used to derive the 5′× 5′ grid map of the % arable land displayed in (d). The resulting mean carbon intensity (CI) of M.×giganteus used as a furnace fuel calculated from this mean yield and the soil organic carbon (SOC) is compared with those of coal is shown as black mask in (c) and areas where the energy-use efficiency (EUE) is less than unity are displayed in (c) as light blue.

The drought kill and frost kill events based upon the criteria for M.×giganteus were calculated and the threshold of one kill events per 15 years, the crop life, set as the cut off for sustainable energy crop production. The mean peak yield for the period 1960–1990 with the cut-off mask for drought kill (brown) and frost kill (blue) is shown in Fig. 2b. The resulting yields in the allowed grid blocks is multiplied by its area and the percent of arable land (Fig. 2d) derived from the CORINE land use map to calculate the net energy yield and carbon mitigation from that block.

This process is repeated for the Miscanthus hi-tech hybrid parameters and the results displayed in a similar manner in Fig. 3 with the exception that we show the interannual variation in yield that can be expected due to meteorological variability by displaying a map of the SD of the dry-matter yield for the period 1960–1990 in Fig. 3d.

Figure 3.

  Miscanthus hi-tech hybrid mean peak dry matter yields for the period 1960–1990 calculated using MISCANFOR (a) compared with the standard deviation (SD) of the mean yield over the same period (d). Figure 2b is the dry-matter yield map with frost kill mask (blue) and drought kill mask (brown) superimposed. The resulting mean carbon intensity (CI) of M. hi-tech hybrid used as a furnace fuel calculated from this mean yield and the soil organic carbon is compared with those of coal is shown as black mask in (c) and areas where the energy-use efficiency (EUE) is less than unity are displayed in (c) as light blue.

The net energy production (NEP) and net carbon mitigation (NCM), considering the difference between burning coal and Miscanthus in the furnace, is calculated for each grid block considering that 10% of the arable land is used. The grid totals are added to the county totals if there is not a kill event and the EUE and CI values favourable. The country values and EU27 total for NEP and NCM are listed in Table 1 for the base case 1960–1990. The energy totals are also converted to barrels of oil equivalent per day to compare with published oil importation and production statistics. The total Miscanthus production on an evenly distributed equivalent of 10% of arable land in EU27 represents 3.6% of 2004 primary energy consumption of 6.5 × 1019 J for the hi-tech hybrid and only half that for M.×giganteus. This would double if we include 10% of pasture as well. Although Miscanthus could be grown on marginal soils, arable land was used to reflect the spatial distribution of excess arable land (under the previous EU set-aside scheme) and land that was available for other uses. In many cases the higher SOC of rough pasture and scrub and woodland or marshy land would preclude its use for growing Miscanthus as the net soil carbon emissions would be high.

Table 1.   Summary of the mean annual net energy yield, energy cost and carbon mitigation for the period 1960–1990 for individual countries and for EU27 using Miscanthus×giganteus as a bio energy crop used as a coal replacement in cofired power stations
CountryNet energy yield (GJ yr−1)Carbon mitigated (Mg C yr−1)EUE (GJ GJ−1)Carbon intensity (g MJ−1)Oil equivalent (bboe day−1)Energy cost (GJ yr−1)
  1. The figures relate to the use of 10% of year 2000 arable land. The annual net energy production is converted to barrels of oil equivalent per day (bboe day−1) to be compared with published energy production and consumption statistics. The carbon intensity of the Miscanthus fuel is listed to compare to coal 33, oil 21 and gas 16 g CO2 C MJ−1. The energy use efficiency (EUE) or energy produced compared with the energy used in its production.

Albania2 238 95569 9764.91.7964453 062
Austria11 648 387297 6444.97.450182 388 808
Belgium27 280 546485 1774.715.211 7525 797 086
Bosnia and Hertzgovina8 585 316235 2455.05.636981 704 366
Bulgaria224 69049474.911.09745 491
Croatia20 200 296595 9075.03.587024 011 598
Czech Republic
Denmark6 439 59245 7894.125.927741 561 834
France328 316 4807 565 5624.810.0141 43068 120 696
Germany109 663 0161 473 2804.719.647 24023 450 346
Greece295 49690564.92.412760 545
Hungary7 725 193212 9885.05.433281 557 807
Ireland12 927 23215 8274.331.855693 003 339
Italy109 678 5283 128 7164.94.547 24722 175 996
Luxemberg2 131 84944 3214.712.2918454 297
Macedonia3 775 39678 4694.912.21626767 311
Netherlands21 182 08850 2464.630.691254 598 160
Portugal11 753 072173 9645.018.250632 350 145
Slovakia2 452 08165 9914.86.11056511 109
Slovenia5 771 416134 7645.09.624861 152 984
Spain58 012 6041 044 7484.915.024 99011 895 940
The United Kingdom89 900 4641 359 9384.317.938 72720 876 338
EU840 202 69717 092 5554.712.7361 938176 937 258

Future climate, land use and technology scenarios. MISCANFOR was run for each of the four SRES scenarios for time slices 2020, 2050 and 2080. Two cases were considered for each scenario-time slice, one using the parameters for M.×giganteus and a second for the hi-tech hybrid. Before running each simulation the mean of the generated 102-year meteorological time series was compared with the CRU scenario time slice value to ensure a unity relationship.

The simulations were run using the same cut offs for EUE and CI as the 1960–1990 simulation. The drought and frost kill cut offs were used for both the M.×giganteus simulation and the hi-tech hybrid (using the best M. sinensis values). For the A2 scenario the mean peak yields for the time slices 2020, 2050 and 2080 are compared with the base case in for M.×giganteus and the hi-tech hybrid. In each case the appropriate parameters for growth and frost and drought kill events are used with a criteria of one kill event per 15 years as a practical economic cut off.

The yield maps are shown in Figs 4 and 5 with the frost and drought kill masks for M.×giganteus and the hi-tech hybrid for each time slice. These maps show an increase in 2020 but a reduction with climate change thereafter. The maps for the hi-tech hybrid provide a more optimistic outlook.

Figure 4.

  Miscanthus×giganteus and Miscanthus hi-tech hybrid mean peak dry matter yields for the A2 scenario time slices shown with a frost and drought kill mask of one events per 15 years for 1960–1990 and A2 2020, Frost kill is shown in blue and drought kill in brown. M. × giganteus (a) 1960–1990, (b) 1960–1990, (c) 2020; hi-tech hybrid, (d) 2020.

Figure 5.

  Miscanthus×giganteus and Miscanthus hi-tech hybrid mean peak dry matter yields for the A2 scenario time slices shown with a frost and drought kill mask of one events per 15 years for A2 2050, A2 2080, frost kill is shown in blue and drought kill in brown. M. × giganteus (a) 2050, (b) 2050, (c) 2080; hi-tech hybrid, (d) 2080.

The energy yields for each country and for EU27 in PJ were compared for each scenario and time slice for both the M.×giganteus and the hi-tech hybrid. The values compared are using 10% of the arable land in 2000 evenly distributed throughout EU27. Net energy yield for each scenario time slice is plotted against year in Fig. 6 and the advance of the summer drought conditions for all scenarios during the decade is reflected in the reduction of the geographical areas for which M.×giganteus is a sustainable crop. In contrast, the frost kill limitation is reduced to Northern Scandinavia by 2080 for all scenarios. The yield of the hi-tech hybrid, in comparison, remains relatively constant as the reduction in yield caused by the drought conditions in the south is balanced by the increase in yields in the north and maritime climates due to warming. The extreme conditions induced by the A1FI and A2 scenarios show a significant yield drop by 2080 even for the new technology hybrid.

Figure 6.

 Energy yields for EU27 countries, using 10% of the 2000 arable land, for the SRES scenarios: A1FI, A2, B1 and B2 for the time slices 2020, 2050 and 2080 compared base case 1960–1990. Two cases are considered, one with the yields limited by the drought and frost kill criteria for Miscanthus×giganteus and another with kill criteria representing a technology improvement to hybrids with yields of M. ×giganteus and a frost and drought tolerant hi-tech hybrid.

Using the area of arable land that was predicted by Rounsevell et al. (2006) to be available for bioenergy crops for the time slices 2020, 2060 and 2080 for each of the SRES scenarios, we calculate the net energy yield and the net carbon mitigated. Because of the frost and drought limits of M.×giganteus, the yield for all scenarios does not rise >3% of EU27's primary energy needs and decreases to <2% for all scenarios by 2080. The hi-tech hybrid is able to produce nearly 12% of EU27's primary energy needs by 2050 for the A2 and B2 scenarios, but rain-fed conditions become difficult for even this hybrid by 2080 for the A1FI and A2 scenario. These energy yields are plotted in Fig. 7, for each time slice for each scenario for M.×giganteus and the hi-tech hybrid.

Figure 7.

 Estimate of the percentage of current (2000) primary energy that can be obtained from Miscanthus for the four future scenarios at three time slices considering both Miscanthus×giganteus and a technically improved hybrid with drought and frost resistance for yields and crop viability range and the land area predicted to be available for bioenergy by Rounsevell et al. (2006).

The yield of each country changes with time as the climate warms and the summer precipitation changes, favouring the maritime and prejudicing the more continental climates. The model produces summaries of the energy yield and the carbon mitigated for each county for each scenario and time slice. Table 2 shows the comparison between the energy yields of each country for M.×giganteus and the theoretical hybrid for the 2080 time slice of the A2 scenario.

Table 2.   Summary of the mean annual net energy yield and carbon mitigation for the SRES A2 scenario for individual countries and for EU27 considering the use of 10% of year 2000 arable land
CountryMiscanthus×giganteusMiscanthus hi tech hybrid
Net energy yield (GJ yr−1)Carbon mitigated (Mg C yr−1)Net energy yield (GJ yr−1)Carbon mitigated (Mg C yr−1)
  1. Two cases are considered: the first is for the Miscanthus×giganteus crop and the second is for a theoretical drought and frost resistant hybrid with the same yield as M.×giganteus. The bioenergy crop is used as a coal replacement in cofired power stations.

Albania  1 707 66451 299
Austria10 811 015294 00326 923 946722 806
Belgium4 184 61424 18432 248 048709 789
Bosnia and Hertzgovina  15 316 307352 376
Bulgaria  00
Croatia560 81815 00310 317 649288 704
Czech Republic4 936 214113 26550 691 4881 213 809
Denmark17 797 984117 72752 900 680586 109
Estonia1 476 08047 61421 510 842276 952
Finland1 724 948574335 121 772−513 265
France2 662 31213 524106 677 2322 447 071
Germany52 200 168597 710195 375 3443 629 120
Greece  00
Hungary  6 062 444219 669
Ireland17 939 416305 09121 470 506380 854
Italy8 633 556242 75532 475 500929 770
Latvia4 168 230129 18435 924 136983 564
Lithuania746 70823 64351 571 9801 454 482
Luxemberg415 38414 4202 140 57252 612
Macedonia  7247133
Malta  00
Netherlands17 042 458309 23128 210 332427 350
Poland63 545 3641 645 774201 871 4244 959 150
Portugal  1 056 41814 254
Romania79 16513888 485 077205 270
Slovakia8 523 397219 26918 907 396480 816
Slovenia7 459 858198 60513 003 219342 316
Spain  22 269 738264 334
Sweden16 746 38943 43644 454 816127 908
The United Kingdom34 131 788765 60681 411 6721 941 581
EU275 785 8665 127 1751 118 113 44922 548 833


In this paper, we have built on the previous studies by further developing a new Miscanthus crop growth model MISCANFOR to make analysis of (i) interannual yield variation due to climate for past and future climates (ii) genotype-specific parameters on yield in Europe. Under recent climatic conditions (1975–2002) we show that 10% of arable land could produce 1709 PJ and mitigate 30 Tg of CO2-C equivalent GHGs compared with EU27 primary energy consumption of 65 568 PJ, emitting 1048 Tg of CO2-C equivalent GHGs in 2005. If we continue to use the clone M.×giganteus, MISCANFOR shows that, as climate change reduces in-season water availability, energy production and carbon mitigation could fall by 80% by 2080 for the A2 scenario. However, because Miscanthus is found in a huge range of climates in Asia, we have no doubt that new hybrids will incorporate genes conferring superior drought and frost tolerance. Using parameters from characterized germplasm, we calculate energy production could increase from present levels by 88% (to 22 360 PJ) and mitigate 42 Tg of CO2-C equivalent using 10% arable land for the 2080 mid-range A2 scenario. This is equivalent to 3.6% of 2005 EU27 primary energy consumption and 4% of CO2 equivalent C GHG emissions. With projections of land available for bioenergy crops we show that Miscanthus could be able to supply 12% of the EU's energy need by 2050. This will be reduced by 2080 for the extreme scenarios such as A1FI and A2.

In detail, the use of MISCANFOR for this study improved the accuracy of Miscanthus yield predictions for future climates in the European area, as shown by Hastings et al. (2009a). In addition the use of a representative meteorological time series for each scenario time slice enabled the interannual variation of predicted energy yields and GHG mitigation to be determined without the distortions created by using mean meteorological conditions (Hastings et al., 2008) that overestimated the yield and created errors in yield distribution by country for EU27. This time series has also enabled the estimation of drought and frost kill events to determine the growing range of two Miscanthus genotypes. Previous studies (Stampfl et al., 2007; Hastings et al., 2008) only considered mean meteorological conditions and gross harvest DM and energy yield.

A recent study (Miguez et al., 2008) developed a meta analysis model to predict yields based upon thermal time, age of plantation, N fertilization and plant density. It also used country and location as random effects in their nonlinear mixed model. Their model considered the variation of yield with plantation age, which is especially useful in the fist establishment years. In our estimation we considered a zero yield for the first year but their methodology could be applied in future work. We modelled the repartition of nutrients to the rhizome and loss of DM from the fall peak yield to the spring harvest. We did not consider autumn harvest as that would require more fertilizer to compensate for the loss of nutrients in the harvest. In addition the field experiments, where an autumn harvest was made (Ercoli et al., 1999; Danalatos et al., 2007) the DM yield was strongly influenced by fertilization. The use of more fertilizer would adversely affect the CI of the biomass fuel. Our study demonstrated the strong dependence of Miscanthus crop on the availability of water and specifically did not consider irrigation due to the high carbon and energy cost as well as the potential shortage of water in the European Mediterranean countries. In the Miguez et al. (2008) study water availability was only implicitly considered in the country/location random variable.

In this study, we have calculated the net energy yield by considering the energy cost of growing and processing the biomass and latent heat of vapourization of the moisture in the biomass. From this we have calculated the net energy yield and EUE. We have also calculated the CI of the fuel considering both the carbon cost of production and the soil GHG emissions. This has enabled a direct comparison to be made with fossil fuel CIs and has enabled us to estimate the net mitigation of GHGs by calculating the difference between using Miscanthus as a furnace fuel in place of coal.

Previous studies (Hastings et al., 2008) considered limiting the Miscanthus crop to arable land due to GHG emissions from high SOC soils under other land management or ecosystems. Here, we calculate the GHG emissions from soil considering its current SOC and as changing land use from pasture, woodland or natural ecosystems to Miscanthus cropping increases GHG emissions to the level that gives a Miscanthus fuel a comparable CI to fossil fuels, this effectively also limits the area available to existing arable land. If we do this, the CI of the Miscanthus fuel is 75% that of natural gas and 36% of the coal it replaces in cofiring. The SOC of long-term arable land is below that of a long-term Miscanthus ecosystem, as reported by Chiang et al. (2004) on a two century Miscanthus ecosystem in the Ta-Ta-Chia area of the Yu-Shan National Park in Taiwan which have an SOC of 98–101 Mg ha−1. This means that as we limit out modelled crop to arable land we predict a net accumulation SOC in some areas (France), which reduces the CI. If Miscanthus would be grown on existing grassland the net accumulation of SOC would be lower or zero and the CI of such Miscanthus bioenergy would be higher. The contribution of SOC changes to the carbon costs of bioenergy is important and requires more experimentation to understand and reduce its impact.

Net energy calculations used the crop management system proposed by DEFRA using the fertilizer additions of potassium, phosphorus and nitrogen that balance the nutrient removal by harvesting the crop in early spring after the nutrient repartition to the rhizomes. We considered the case that 100% of the crop was used remotely at a mean distance of 300 km, reflecting cofiring at a facility such as DRAX. This gave a fixed energy cost of 9.07 GJ ha−1 yr−1 and a yield-related cost of 2.34 GJ Mg−1 yr−1. If the optimum model of local use of the crop rather than cofiring in large power stations, and if the crop were propagated by rhizome, then the cost would drop to a fixed energy cost of 5.64 GJ ha−1 yr−1 and a yield-related cost of 0.61 GJ Mg−1 yr−1. If Miscanthus could be propagated by seed the fixed cost would be reduced to 2.6 GJ ha−1. From this it can be seen that the EUE and CI could be greatly improved by advances in breeding technology.

CI will be improved for the same reason if the transportation of the biomass is minimized by local use (Smith & Smith, 2000). In addition, it is possible that warming climates and dryer summers, with appropriate genotype selection, could radically alter the harvest time from spring to late summer. Such a change would have impacts on carbon mitigation and energy yield. In our scenarios, we considered planting using micropropagation for the main simulation and the results presented here, but other options such as seed and rhizome planting improved the overall EUE and CI but did not impact the overall yield significantly. In the model, we used a 3-year establishment period. This is a simplification as experimental plot anecdotal evidence indicates that 2 years in the south is fine, 3 years in the central area and 4 years in cold areas is more appropriate, but there in insufficient data to effectively parameterize this thermal effect. Some crop trials have shown that there is a relationship between establishment and plant spacing and some have achieved establishment yields in 2 years. If the crop can be propagated by seed or the lifetime of the crop be extended, this will reduce the overall carbon and energy cost. These factors should be accorded priority in research to improve Miscanthus cultivars and be included in future model developments.

Previous work to determine the spatial range of a sustainable Miscanthus crop did this either by minimum yield (Stampfl et al., 2007) or SOC and minimum yield (Hastings et al., 2008). Tuck et al. (2006) used the proxy criteria of maximum and minimum elevation, summer temperature and annual precipitation to determine the range, or bioclimatic envelopes, of various bioenergy crops including Miscanthus for four climate scenarios and four climate models for 2020, 2050 and 2080 time slices. We show with this study that for a perennial crop such as Miscanthus, interannual variation is an important consideration due to the 15–20-year crop life, because crop kill by extreme frost or drought events at any stage during the life cycle of the crop necessitates replanting, which make Miscanthus growth at that location unsustainable.

This study demonstrates that winter minimum temperature and the summer precipitation are the critical explanatory variables to determine the sustainability of the M.×giganteus bioenergy crop, and in particular, their interannual variability. We show that with the annual progression of all climate scenarios, although the winter warming migrates the sustainability limit northwards to areas where historically there is currently not much arable land, most of the areas of Europe with the most arable land become increasingly hostile environments for M.×giganteus, as events when the soil water falls below wilt point for >60 days become more frequent due to higher evapotranspiration and lower summer precipitation. If Miscanthus varieties could be developed that grow in short days as a winter crop then larger yields could be achieved in the Mediterranean area when winter temperatures are >10 °C and soil moisture is at its highest, however, it would have to be harvested after the nutrients have been repartitioned to the rhizome to minimize fertilizer use and maximize DM yield. Based upon observations from field experiments in Italy, Spain, Portugal, Turkey and Italy (Hastings et al., 2009a), we have assumed in the model that Miscanthus has a photoperiod sensitivity by using day length to control the length of the growing season. This requires experimental verification.

In this study, we considered growing Miscanthus on excess arable land for two reasons. The fist is that it gave a geographical distribution of where Miscanthus could be planted, and secondly, that previous studies of the production of bioenergy crops had considered using first the 10% of arable land that was EU set aside to avoid the excess production of food and secondly the excess arable land that would be available when the EU accession countries crop yields reach the EU average by applying modern farming techniques. We have not accounted for future changes in the distribution of arable land with climate change as these would entail a land use change with the corresponding emissions of GHG. In general, there will be a net SOC gain when growing Miscanthus on arable land but if Miscanthus is grown on rough pasture, scrubland, woodland or wetlands there could be a net SOC loss, which would result in a CI of the fuel that is greater than coal. However, future work should consider the conversion of pasture or degraded land to Miscanthus cropping using zero/low till techniques but to avoid net soil carbon loss the marginal land should have an SOC lower than or equal to the Miscanthus SOC equilibrium level. However, there is no published data on no-till planting options for Miscanthus and this requires future field trials. Water availability strongly affects where Miscanthus can be grown for future climate scenarios and this will also affect food crops and will increase the competition for land use between food and fuel.

Clifton-Brown & Lewandowski (2000b), Clifton-Brown et al. (2002) showed that some genotypes of M. sinensis had a better drought and frost tolerance than the widely trialled M.×giganteus (Clifton-Brown & Lewandowski, 2000a), Although there is some indication from trials that the frost and drought tolerance increases with the life of the plant, it has not been quantified experimentally, so we have not considered the effect of hardening or acclimatization to increase frost tolerance during the life of a crop.

Our simulation of a drought and frost tolerant theoretical hi-tech hybrid with the yield of M.×giganteus projects the upper limit of the energy yield and carbon mitigation that would be possible with this energy crop and predicts that 12.6% of EU27 primary energy is the ceiling for biomass energy. However, this energy has a relatively low CI, as we determine that for the period 1960–1990 the mean CI of the Miscanthus fuel would be 12.7 g CO2 C MJ−1, which compared favourably to gas, which has a CI of 16 g CO2 C MJ−1. The CI does not significantly change as the climate warms, unless we grow Miscanthus on land other than arable with a higher SOC (such as pasture, woodland, peat land or other natural ecosystem land). We also demonstrate that breeding research, to develop high yielding, water efficient Miscanthus hybrids, is vital to achieve even this level of bioenergy production. In our estimations of carbon mitigated we assumed the cofiring scenario, eliminating some coal use, if Miscanthus replaces gas then the mitigation will be less.

In this study, we did not consider other bioenergy crops, but other studies have indicated that Miscanthus has potentially the highest energy yield and EUE of all currently considered species (Sims et al., 2006). Other perennial energy crops such as short rotation coppice are C3 plants that have a lower photosynthesis rate to the C4 crops that use 40% less N and water (Monteith, 1978) so will have a lower energy yield and higher energy cost. Of the perennial C4 plants trialled as a bioenergy crops, Miscanthus has be shown to have higher yields than others like switchgrass and reed canary grass (Heaton et al., 2008). All perennial crops will suffer from the same summer drought conditions, and even though they do not occur each year may make their cropping unsustainable. For annual crops such as those grown for both food and energy, like wheat and maize, it is clear that they will also suffer increasing crop failures and low yields due to summer droughts (Parry et al., 2005). However, with the potential to provide 12% of the EU's energy needs, in the B2 scenario, by the year 2050 Miscanthus bioenergy could be an important contributor to reducing the EU's GHG emissions by up to 14% whilst providing sustainable energy. We conclude that even if a mixture of energy crops is grown in future, these predicted Miscanthus yields show the upper limit of bioenergy production that is possible in the EU27 using the available land proposed by Rounsevell et al. (2006).


Bioenergy has the potential to reduce Europe's GHG emissions by up to 14% while providing sustainable energy. Miscanthus could play a significant role in the renewable energy mix to provide 12% of Europe's primary energy needs, in the A2 scenario, by the year 2050 by using 35% of current arable land. Interannual variations in yield could affect supply by ±25% if we use M.×giganteus but if we develop the new tech hybrid then this is reduces to ±20%. To achieve this potential M.×giganteus must be replaced by more drought and frost resistant types to future proof energy crop yields. Changes in agronomic methods could double energy output–input ratios and reduce the CI.


This work was funded by a Sixth Century Scholarship from the University of Aberdeen as a joint project between the College of Physical Sciences and the College of Life Sciences and Medicine. We thank Prof. Mike Jones of Trinity College Dublin for suggestions during the field work of Cashel used in the analysis and Dr Uffe Jorgenson and Dr Klaus Hammel for early discussions on the model. Data were derived from numerous projects, of particular note are the EU contracts FAIR3-CT96 –1392 (EMI), FAIR3-CT96-1707 (EMN) and SSPE-CT-2005-006581 (ENFA). Other trial data was taken from publications and is acknowledged in the text. P. S. is a Royal Society-Wolfson Research Merit Award holder.