• Open Access

Biomass production potential from Populus short rotation systems in Romania


  • Christian Werner,

    Corresponding author
    1. Biodiversity and Climate Research Centre (BiK-F), Senckenberg Gesellschaft für Naturforschung, Frankfurt, Germany
    Search for more papers by this author
  • Edwin Haas,

    1. Institute for Meteorology and Climate Research, Atmospheric Environmental Research Division, Institute for Atmospheric Environmental Research (IMK-IFU), Karlsruhe Institute of Technology, Garmisch-Partenkirchen, Germany
    Search for more papers by this author
  • Rüdiger Grote,

    1. Institute for Meteorology and Climate Research, Atmospheric Environmental Research Division, Institute for Atmospheric Environmental Research (IMK-IFU), Karlsruhe Institute of Technology, Garmisch-Partenkirchen, Germany
    Search for more papers by this author
  • Martin Gauder,

    1. Institute of Crop Science, University of Hohenheim, Stuttgart, Germany
    Search for more papers by this author
  • Simone Graeff-Hönninger,

    1. Institute of Crop Science, University of Hohenheim, Stuttgart, Germany
    Search for more papers by this author
  • Wilhelm Claupein,

    1. Institute of Crop Science, University of Hohenheim, Stuttgart, Germany
    Search for more papers by this author
  • Klaus Butterbach-Bahl

    1. Institute for Meteorology and Climate Research, Atmospheric Environmental Research Division, Institute for Atmospheric Environmental Research (IMK-IFU), Karlsruhe Institute of Technology, Garmisch-Partenkirchen, Germany
    Search for more papers by this author


The aim of this study was to assess the potential of biomass production by short rotation poplar in Romania without constraining agricultural food production. Located in the eastern part of Europe, Romania provides substantial land resources suitable for bioenergy production. The process-oriented biogeochemical model Landscape DNDC was used in conjunction with the forest-growth model PSIM to simulate the yield of poplar grown in short-rotation coppice at different sites in Romania. The model was validated on five sites with different climatic conditions in Central Europe. Using regional site conditions, with climatic parameters and organic carbon content in soil being the most important, the biomass production potential of poplar plantations was simulated for agricultural areas across Romania.

Results indicated a mean productivity of 12.2 ± 0.5 t ha−1 year−1 of poplar coppices on arable land in Romania. The highest yields were simulated for lowland areas in the south-east and west and for the Mures valley, whereas the lowest yields – due to either temperature or water limitations – were found for the mountainous regions, the Danube valley, and the region west of Bucharest. The amount of abandoned arable land in the past 10 years indicates that around 10% of cropping land in production in 1999 (approximately 1 million ha) is available for bioenergy production systems today. Production of poplar grown in short-rotation coppices on these areas would result in a yield of approximately 10 million tons of wood per year. The energy that can be generated by conversion of poplar short rotation coppice biomass may contribute up to approximately 8% of the national energy demand if these set-aside areas are used for lignocellulosic bioenergy.


Establishing energy security and combating climate change have become important foci of policies for the vast majority of governments around the world. The European Union (EU) has set binding targets to increase energy security while at the same time reducing greenhouse gas emissions. This policy requires member states to increase their use of renewable energy sources to a share of 20% of the overall use of energy by 2020 (European Commission, 2009).

Biomass as a source of energy is regarded as one important component in the renewable energy mix. Regarding biofuels, biodiesel, and ethanol are the most important commodities, with the EU being the most important global biodiesel market (Lamers et al., 2011). However, first generation biofuels are often associated with an unfavorable energy and greenhouse gas balance and are moreover discussed controversially regarding a possible competition with food crops (Lamers et al., 2011; Crutzen et al., 2008; Gauder et al., 2011; Tenenbaum, 2008; Walker, 2009). Biomass from cellulosic bioenergy crops is expected to play a substantial role in future energy systems, especially if climate policy aims to stabilize greenhouse gas concentration at low levels (European Commission, 2009; Lamers et al., 2011; Popp et al., 2011; Crutzen et al., 2008; Gauder et al., 2011; Tenenbaum, 2008; Walker, 2009).

Compared to annual agricultural products, woody biomass production from perennial crops in short rotation coppice (SRC) has several environmental advantages such as a higher retention of nitrogen (Lamers et al., 2011; Hellebrand et al., 2010; Kern et al., 2010), positive impacts on soil ecology (Lamers et al., 2011; Kahle et al., 2010; Crutzen et al., 2008; Gauder et al., 2011; Tenenbaum, 2008; Walker, 2009) and on biological diversity (Riffell et al., 2011; Bremer & Farley, 2010; Hartmann et al., 2010; Rowe et al., 2009). Many of the effects are due to fewer requirements for herbicide and fertilizer application (e.g., Hellebrand et al., 2010). SRC has been shown to sequester carbon above- and belowground (Rytter, 2012; Hellebrand et al., 2010; Kahle et al., 2010; Hillier et al., 2009) and, given appropriate management, this can be sustained over time (Nassi O di Nasso et al., 2010; Pellegrino et al., 2011). SRC can also be used for re-cultivation of former mining sites (e.g., Bungart & Hüttl, 2004) or for the restoration of marginal lands which are not suitable for food production (Williams et al., 2009). Overall, for many European regions SRC seems a suitable commodity for biofuel production, with poplar being the most promising species (Tullus et al., 2012; Djomo et al., 2010). Hybrid poplars typically reach a mean annual growth increment of 6–16 t dry matter ha−1 year−1 throughout their rotation period, usually with increasing growth rates throughout the first 3–6 years and maximum reported yields of about 35 t ha−1 year−1 (Boelcke & Kahle, 2008). Considering future climatic conditions and an increased CO2 air concentration, the area of suitable sites will increase over most regions in Europe with only few areas getting too dry to be productive anymore (Bellarby et al., 2010; Christersson, 2010; Kollas et al., 2009; Christersson, 2010; Liberloo et al., 2010). Due to the high growth potential of SRC species, they are assumed to be a suitable substitute for fossil fuels (Baral & Guha, 2004; Popp et al., 2011). Wood chips derived from SRC can be burned directly or converted to liquid fuels. Biomass-to-liquid (BTL) and bioethanol production from cellulosic feedstock are the most promising techniques to convert woody biomass to convenient biofuels, although the production processes are not yet fully commercialized.

Although growth potentials are traditionally estimated with empirical models (Benbrahim et al., 2000; Von Röhle et al., 2006), a substantial increase in application of SRC in combination with expected climate changes limits this approach. Since a low input of fertilizer and irrigation is central to a concept that aims to increase the carbon sequestration potential, local climatic and soil conditions are the determining factors. However, with respect to the large targeted areas that might be suitable for SRC, the environmental combinations are exceeding the range that could be covered with existing empirical growth relations. Phenological changes due to higher spring temperatures and accompanying nutrient requirements as well as the interrelations between root allocation and belowground resource availability were identified as critical issues that could not be covered in empirical models (Karp & Shield, 2008).

It took considerable efforts to finally develop process-based models into full ecosystem models that where applicable for long-term simulations since these are computational demanding and sensitive to uncertain initializations (Battaglia & Sands, 1998; Johnsen et al., 2001; Landsberg, 2003; Mäkelä et al., 2000). On the other hand, the implicit assumption of most process-oriented biogeochemical models that the forest is homogenous and can be represented by one average tree is best met in forest plantations. Thus, process-oriented models are increasingly applied to large regions following an a priori parameterization and evaluation of model results on a set of experiments (Miehle et al., 2006, 2010). A very prominent example is the 3PG model that has been applied for various plantations all over the world (e.g., Rodriguez et al., 2009; Zhao et al., 2009; Amichev et al., 2010). To our knowledge, the first process-oriented model that has been applied on SRC was the SECRETS model (Deckmyn et al., 2004). Other examples for applications on poplar plantations over a large range of conditions have been executed with the models ORCHIDEE, Biome-BGC, and 4C (Liberloo et al., 2010; Lasch et al., 2010). However, the importance of soil processes has thus far only been acknowledged in few model simulations (Luxmoore et al., 2008).

We employed a process-based model with a strong biogeochemical component to investigate the impact of climatic as well as soil-related site properties on potential poplar biomass production for different regions in Romania. The model used is a derivative of the DeNitrification-DeComposition (DNDC) and Forest-DNDC/PnET-N-DNDC models, which have been widely applied for estimating soil greenhouse gas emissions in Europe, Australia and other regions worldwide (e.g., Kesik et al., 2005; Kiese et al., 2005; Werner et al., 2007; Miehle et al., 2006; Butterbach-Bahl et al., 2009). Divergent process descriptions and implementations between the model varieties were unified in a recent restructuring approach which led to the development of the MoBiLE modeling framework (e.g., Grote et al., 2011). Furthermore, new sub-modules were incorporated into this framework to provide alternative/improved model descriptions for specific processes (e.g. soil hydrology) or ecosystem representations (plant growth). For instance, an alternative forest growth-model (PSIM), which better considers foliage dynamics and dimensional growth and calculates light conditions in dependence on stand density, was included (Grote, 2007; Grote et al., 2011). It features dynamic phenology based on growing degree-days, demand-driven (leaf mass) and photosynthetically controlled root allocation rates and a photosynthesis description based on light intensity, temperature, ambient CO2 concentration, and leaf development stage. Nitrogen uptake is controlled by plant demand (given by optimum N concentration in tissue), specific uptake capacity and available extractable nitrogen in the rooting zone (Grote et al., 2009). The PSIM vegetation sub-module, as any other sub-module in the framework, can be selected at runtime and was used in conjunction with the DNDC sub-modules for microclimate, soilchemistry and water cycling. The chosen model framework setup (PSIM + DNDC) was previously evaluated for young poplar development (Behnke et al., 2012) and is thus used here to calculate biomass development of poplar plantations. Key PSIM model parameters determining gas exchange and physiology of poplar are given in Table 1.

Table 1. Important parameters describing the gas exchange and growth of poplar trees in the Landscape DNDC (vegetation module: PSIM) model. Gas exchange parameters refer to the Farquhar et al. (1982) and the Ball et al. (1987) model description. In addition, values for activation energies are taken from Zhu et al. (2010). Growth parameters are based on the allocation concept presented in Grote (1998). Additional parameters are used to describe the dynamic relationship between sapwood area at 1.3 m and foliage leaf area (according to Ewers et al., 2005) as well as to distribute stemwood growth into height and diameter increase (from Honer, 1967)
Allometric parameters 
 Foliage biomass at optimal,closed canopy conditionMFOLOPTkg m−20.48Deckmyn et al. (2004)
 Ratio between fine root- and foliage biomass at standard conditionsQRF 0.26Giradin et al. (2008)
 Below ground wood fraction UGWDF0 … 10.3Behnke et al. (2012)
 Wood densityDSAPkg DW dm−30.41Pliura et al. (2007)
 Specific leaf area (full light)SLAMAXm2 kg−115.7Calfapietra et al. (2005)
 Specific leaf area (shade) SLAMIN 10.8Calfapietra et al. (2005)
Gas exchange parameters 
 Max. RubP saturated rate of carbo-xylation at 25 °C for sun leavesVCMAX25μmol m−2 s−1117.0Silim et al. (2010)
 Max. electron transport rate at25 °C for sun leaves JMAX25μmol m−2 s−1216.0Silim et al. (2010)
 Slope of foliage conductivity in response to assimilation in the BERRY-BALL modelSLOPE_GSA 11.3Zhu et al. (2010)
 Rel. available soil water content at which stomata conductance is affectedH2OREF_A0 … 10.3Brilli et al. (2007)
 Maximum stomata conductivityGSMAXmmol H2O m−2 s−1800.0Ethier & Livingston (2004)

The current model framework (now called Landscape DNDC) also allows for true coupling of heterogeneous land-use and soil-vegetation processes in a landscape with other 2D models and full bi-direction information exchange (Baur et al., 2006; Migliavacca et al., 2009; Haas et al., 2012). Note that although the codebase was restructured and expanded to allow for the new features, the model still inherits all process descriptions and model properties from the previous versions.

In this study, the potential growth and biomass yield of poplar grown in a short rotation system was simulated for agricultural land across Romania. Located in the eastern part of Europe, Romania provides substantial land resources which are at present not used for agricultural production or are being abandoned due to unfavorable environmental site conditions and/or unviable economic benefits (Reif et al., 2008; Grote et al., 2011; Baumann et al., 2011; Haas et al., 2012; Eurostat, 2009; Müller et al., 2009).

The aim of this study was to evaluate the potential for poplar plantations on different sites in Romania. Furthermore, potential impacts on agricultural performance by the implementation of poplar coppices are discussed.

Materials and methods

Model validation

Physiological properties relating to gas exchange, early growth and senescence were evaluated with data from a greenhouse study that followed the development of young hybrid poplar during a 2-year investigation period (Behnke et al., 2012). For field scale evaluation of gas exchange we further evaluated the model with Eddy-Covariance data from the Euroflux site Parco Ticino in Italy (Migliavacca et al., 2009). Results from this particular exercise are not shown but similar evaluations have previously been presented for other major European tree species (Grote et al., 2011). To demonstrate that the yield from typical short-rotation forestry with poplar could be simulated realistically accounting for dependence on environmental conditions, we applied the model to five European sites where biomass yield was available from literature sources (Table 2). These sites were initialized with rudimentary soil information as specified in the references and run with climate data derived from the NitroEurope project (http://www.nitroeurope.eu). Simulations were run for as many years as evaluation data were available, which was either 5 years [Bavaria (BAV), Thuringia (THUR), Czech Republic 2 (CZE2)] or 6 years [Mecklenburg-Western Pomerania (MWP), Czech Republic 1 (CZE1)]. Results for total simulated vs. measured aboveground biomass over the simulation period are presented in Fig. 1. The comparison shows that the model parameterization gives reasonable biomass projections for the different sites with a slope of 1.08 and an RMSE of 2.3 [t ha−1 year−1] in average over all sites. Similar good results were also obtained with regard to other plant parameters, specifically dominant height and mean diameter at breast height (data not shown).

Figure 1.

Measured and simulated aboveground biomass yield for three sites in Germany (THUR: Thuringia, MWP: Mecklenburg-Western Pomerania, BAV: Bavaria) and two sites in the Czech Republic (CZE1, CZE2). Growth values are average annual values over 5–6 years (see more details in text and Table 2).

Table 2. Field sites with Poplar clones ‘Max’ used for model validation (MAT: mean annual temperature [°C])
SiteCountryRegionDistrictLatLonMATSoil typeReferences
THURGermanyThuringiaDornburg51.011.78.8loess loamBiertümpfel et al. (2009)
MWPGermanyMecklenburg-W. Pom.Gülzow53.812.18.2cambisolBoelcke & Kahle (2008)
BAVGermanyBavariaBeuerberg47.811.47.5sandy loamBurger (2010)
CZE1Czech R.Moravian-highlandDomanínek49.516.37.3luv. cambisolTrnka et al. (2008)
CZE2Czech R.Central BohemiaPeklov50.114.29.1Weger (2009)

Driving data

Detailed process-based models (such as members of the DNDC model family) require a set of detailed driving data for operation. In this study we used the European modeling database developed within the NitroEurope project (http://www.nitroeurope.eu). The database consists of soil, land-use, climate, and management information (Table 3). A selection of input parameters and their spatial characteristics in the simulation domain are given in Fig. 2. For this study we extracted the NitroEurope Calculation Units (NCUs), spatial units of land with common soil or climatic conditions, comprising Romania from the full NitroEurope spatial database (data source: http://afoludata.jrc.ec.europa.eu/index.php/experiment/detail/2, for Romania arable land: n = 1635). Since we investigate the potential for conversion of arable land to Populus plantations, we only consider NCUs containing arable land (areas with no arable use are thus not considered in the provided maps, Fig. 2).

Table 3. List of input datasets used as model drivers for simulating biomass yield from poplar in Romania. The data for this study were extracted from the NitroEurope modeling database (http://afoludata.jrc.ec.europa.eu/index.php/experiment/detail/2)
Spatial unitsNANitroEurope Calculation Units (NCUs). NCUs share the same administrative unit, the same soil (SGDB classification), the same slope data (CCM DEM 250 in five classes) and a similar difference from the DEM height
LanduseNACorine Land Cover for NCU. CLC2000 resampled at 1x1 km and reclassified
Soilorganic carbon, clay, sand, pH, bulk density, thicknessSoil properties at all NCUs. Spatial averages for A, B and C horizons obtained using geostatistical cokriging approach. Observations derived from the ISRIC WISE/SPADE dataset, landuse specific (arable in this study).
WeatherDaily min./max. temperature, precipitationWeather data on NCUs. Generated from MARS 50 km daily and CRU/ATEAM 0.1deg monthly data.
Figure 2.

Regional site, soil and climatic characteristics of the model domain (from top left to bottom right: elevation [m], annual precipitation [mm year−1], average annual temperature [°C], pH, clay content [%] and carbon content [%]; all soil data given for top soil layers 0–30 cm).

As shown in Fig. 2, the western, eastern, and southern regions of Romania are classified as arable land at low elevation (<300 m a.s.l.) with either low (500–650 mm in the west and south) or very low (<400 mm) average annual rainfall, with latter regions being predominantly found in the east and especially in the Dobrogea province bordering the Black Sea. Higher precipitation amounts are present in the central highland (Carpathian Mountains), although not exceeding 1000 mm. Average annual temperatures follow topographic gradients with lowest annual temperatures fall to 4–5 °C in the mountainous regions, whereas average annual temperatures in the lowlands may be up to 13 °C. Also worth mentioning is the continentality of climatic conditions in Romania, with dry and cold conditions in the winter, prevailing dry and hot conditions during summer and rainfalls occurring mainly in spring and autumn seasons. Topsoil (0–30 cm) conditions also vary substantially over the entire country with high pH (>7) encountered in the eastern and western lowlands and low pH in the mountainous regions. Clay content is low (20–30%) in most areas with small areas of higher clay content in the southern parts. Soil organic carbon in topsoil is low (<2%) for the majority of areas.

Soil data were provided for all spatial computation units (stratified into three general strata sections A, B, and C) according to the NitroEurope soil data (see Table 3). For each profile section soil attribute data for organic carbon, clay content, soil pH, bulk density, and thickness were provided and each strata section was further subdivided into 2 cm (A), 5 cm (B), and 10cm (C) layers. No initial litter layer was prescribed. Poplar saplings were initiated with 10,000 saplings per ha and 0.5 m initial plant height. As required by DNDC models, daily minimum and maximum temperature and precipitation data were provided. Annual nitrogen deposition was applied as wet deposition (e.g., during rainfall events). No additional nitrogen fertilization was implemented.

Processing setup

To limit short-term climatic effects on the simulation outcome, we simulated biomass yields from short-rotation (6 year) Populus plantations for three time slices (1990–1995, 2000–2005, and 2010–2015) with climate data taken from the NitroEurope database and provided the simulated yield averages and standard deviations of these runs. The time step for all model processes is daily, except for photosynthesis which is calculated hourly using temperature and solar radiation values derived from daily data. The same model setup was used for site simulations during the validation phase. For efficient computation, the simulation was performed on a Linux cluster system at IMK-IFU, Garmisch-Partenkirchen, and the simulation domain was split according to available computing nodes (n = 108). Pre- and post-processing tools developed within the NitroEurope IP were used to handle to input data and to assemble model results.

Spatial analysis

To investigate the regional potential for biomass production, we introduce a so-called P-index which describes the potential for biomass production at a given location. The index incorporates three crucial determinants of plant growth a) temperature over the vegetation period (expressed as growing degree-days GDD), b) precipitation, and c) soil fertility, with last being estimated from soil organic carbon concentrations. Note that Landscape DNDC uses fixed C : N-ratios for calculating nitrogen content from organic carbon content if nitrogen is not provided as an input variable.

GDD was calculated as:

display math

where: Tbase = 10 °C (base temperature for arable crops), Tmax = 30 °C.

Only positive GDDi values were considered and the final GDD value used was the average GDD for the years 1971–2000.

We use Tbase = 10 °C (a default for crops) as we are operating on site of arable land use and thus want to consider high and low productivity potential with regard to arable production. Note that this is only used for classification of regional properties and not a model parameter.

Locations were ranked for all three variables (low: less suited for biomass production, high: well suited for biomass production). Finally, the individual ranks were normalized and summed to create a potential productivity index or P-index.

Thus, the P-index is calculated as follows:

display math

where: Precip = average annual precipitation, Corg = organic carbon content in topsoil.

Max is maximum value of a given parameter for all calculation units located in Romania.

All plots were produced using the R software package (R Development Core Team, 2011). All biomass yield data from regional simulations are reported as the average annual yield as dry matter (DM) for the three simulated time slices (standard deviation of these results are given if appropriate). The simulated yield data of all NCUs were area-weighted prior to this averaging.


Figure 3 shows the spatial characteristics of the three input variables and the regional distribution of the resulting P-index. Based on the P-index the agricultural production area is divided into six classes ranging from low potential (LP) areas to high potential (HP) areas (see Table 4, Fig. 4). Cumulative biomass production results are also reported in Table 4. Realistically, biomass production will only occur on marginal land (low suitability for arable production, low P-Index, i.e. LP10, LP25, LP50) so the data for HP10, HP25, and HP50 areas are mainly provided as a reference and for discussion.

Figure 3.

Identification of low potential (LP) and high potential (HP) biomass production regions. The factors average annual precipitation (precip, upper left panel), growing degree-days (GDD, upper right panel), and organic carbon content of top soil (Corg, lower left panel) were ranked individually with scores ranging from 0 (low potential) to 1.0 (high potential) and finally combined and normalized (total, lower right panel).

Figure 4.

Location of production potential rank classes as derived by the P-index (low: −−− high: +++).

Table 4. Simulated biomass production for high potential (HP) and low potential (LP) regions. Given are the average biomass production and cumulative sums for the 10%, 25%, and 50% lowest and highest potential regions as identified by the P-Index (SD: standard deviation of the three modelled time slices 1990–1995, 2000–2005, 2010–2015)
RegionRankYield [t ha−1 year−1]Area [1000 ha]Cumulative yield [million t year−1]
Low potential regions
LP10−−−10.63 (0.46)993.49.92 (0.43)
LP25−−−, −−10.78 (0.39)2,369.425.55 (0.92)
LP50−−−, −−, −11.33 (0.39)4,761.353.96 (1.85)
High potential regions
HP10+++13.70 (0.50)987.813.54 (0.50)
HP25+++, ++13.22 (0.46)2,428.432.11 (1.11)
HP50+++, ++, +13.05 (0.52)4,765.462.18 (2.49)
Total 12.19 (0.46)9,526.7116.4 (4.3)

The simulated area-weighted mean potential annual biomass production by poplar short-rotation plantations on arable land in Romania was 12.2 ± 0.5 t ha−1 year−1 (min: 6.5 t ha−1 year−1, max: 14.8 t ha−1 year−1). The regional differences were pronounced, with the highest yields being simulated for lowland areas in the South-East and West and also the Mures valley. Lowest yields were obtained for the mountainous regions, the Danube valley, and the region north-west of Bucharest (Fig. 5). The simulated biomass production rates matched reported field measurements for sites in Central Europe well (Table 2, Fig. 1). However, poplar biomass production potentials were significantly affected by inter-annual climate variability and were moreover depending on the chosen time period (Fig. 6). Figure 6 further illustrates that yields of 11 t ha−1 year−1 and 13.5 t ha−1 year−1 are dominant, essentially indicating a bi-modal distribution of large areas with differing biomass production potentials. The regional pattern shows good agreement with the postulated potential index rank map (Fig. 3).

Figure 5.

Average simulated biomass yield [t ha−1 year−1]. Given are the average results for the simulated time periods (1990–1995, 2000–2005, 2010–2015).

Figure 6.

Density plot of area-weighted annual biomass production for the simulated spatial entities (orange: 1990–1995, red: 2000–2005, purple: 2010–2015, black: average of three simulations, units: t ha−1 year−1).

Assuming that all arable land in Romania is used for poplar biomass production, the total potential yearly biomass production would amount 116.1 ± 4.3 million t (Table 4). Obviously this scenario is impossible, so the comparison of yields from high and low production (HP and LP) areas with differing areal extension is more meaningful. If, for instance, one limits poplar biomass production on the 10% land with the lowest agricultural productivity (LP10) the average biomass yield would be 10.6 ± 0.5 t ha−1 year−1. By including 25% of land with lowest agricultural productivity (LP25) the average biomass yield slightly increases to 10.8 ± 0.4 t ha−1 year−1 as now also sites with better P-index are included. In contrast, if one would be willing to sacrifice 10% of the land with highest P-index or potential agricultural productivity, the average biomass yield would increase by 36.5% as compared to the LP10 scenario (see Fig. 7 for comparison of total LP and HP yields and the relative differences).

Figure 7.

Comparison between cumulative potential biomass production for high (HP) (gray) vs, low (LP) potential regions (white) assuming the use of either 10 (a), 25 (b) or 50% (c) of total arable land with high or low production potentials. The percentages given in the figure state the yield difference between HP and LP sites (e.g., HP10 vs LP10, …).


In Romania, the amount of land used for agricultural practices was reduced by one million ha between 1999 and 2009 (Eurostat, 2011). This reduction was mainly due to decreasing the amount of arable land from 9.4 million ha in 1999 to 8.8 million ha in 2009, while utilization of grassland showed a similar trend with a reduction of 0.46 million ha or 9.6%. These areas were abandoned and have not yet been transformed into forest areas since forest area remained stable at about 6.7 million ha. Abandoned agricultural land is mostly found on marginal sites such as mountainous regions (Baur et al., 2006), which do not offer satisfying yield capacities for agricultural production. Another driver for land abandonment is higher incomes outside the agricultural sector (Reif et al., 2008; Baumann et al., 2011). The exact amount of fallow land is hard to evaluate since arable areas which were abandoned more than 5 years are not recorded as fallow and green manures by Eurostat and referred to as other land (European Commission, 2002). However, according to national statistics 1.03 million ha arable- and grassland have been taken out of production during the past 10 years in Romania (Eurostat, 2011). The percentage of abandoned agricultural land in Romania is comparable with other East European countries, which in some cases have even larger areas of abandoned farmland. Approximately 30% of farmland in Western Ukraine used during the time of the Soviet Union was abandoned after its breakup (Baumann et al., 2011; Kuemmerle et al., 2010). A high share of abandoned land (21%) was also reported for Bulgaria (Vranken et al., 2011). According to the FAO small Balkan countries which suffer from a troubled history (e.g., Bosnia-Herzegowina, Albania) also have vast resources of fallow agricultural land (Faostat, 2012; Witmer & O'Loughlin, 2009).

Assuming the abandoned land in Romania is mainly located on sites with the lowest yield capacities, as most of the literature suggests, the amount and location of these areas resemble the 10% of low productivity areas (LP10) identified in this study. Establishment of poplar plantations on these sites would yield approximately 10 t ha−1 year−1 or a total of 10.6 million t of poplar wood per year for all areas. Since poplar wood has a heating value of 18.4 GJ t−1, total energy harvested would account for 194.3 PJ. This energy value is equivalent to 17.5% of Romania's total national energy consumption (International Energy Agency, 2011). However, losses during storage and combustion or conversion efficiency will decrease the amount of energy gained as electricity, heat or alternative fuel. Energy efficiencies of conversion processes, such as wood to electricity or wood to Fischer-Tropsch Diesel, are between 0.3 and 0.5 (Patzek & Pimentel, 2005). The results indicate that a significant contribution of poplar plantations to the national energy demand is possible by utilizing these abandoned agricultural areas for short-coppice rotation systems. After conversion between 5.3% and 8.8% of Romania's total energy demand could be supplied on basis of fast growing tree plantations without interfering with conventional crop production. These numbers are slightly larger as the outcome from a study on SRC plantation potential for England where the authors found that the introduction of SRC plantations on marginal land could cover approximately 4% of the total energy demand of the United Kingdom (Aylott et al., 2010).

By utilizing the less productive half of all agricultural sites in Romania for biomass plantation it would be possible to harvest wood, which resembles 89.5% of the total energy consumption. This is rather significant since use of woodchips for energy production would reduce significantly total national greenhouse gas emissions. Based on a review of existing literature, Djomo et al. (2010) estimated that short rotation woody crops yield 14.1–85.9 times more energy than coal (energy ratio of coal–0.9) per unit of fossil energy input, while greenhouse gas (GHG) emissions were 9–161 times lower than those of coal. A life cycle analysis for bioenergy crops in England and Wales showed that a conversion from arable land or grassland to SRC poplar plantation was beneficial to the GHG balance even before fossil fuel displacement was taken into account and even more though than other bioenergy crops (Hillier et al., 2009).

On the other hand, the utilization of not only the currently abandoned agricultural land but of half of the arable land for poplar plantations would likely constrain food production in Romania. This would still be the case even if one assumes that the agricultural production efficiency in Romania, which is rather low at present (Eurostat, 2009) increases due to land-use intensification and use of modern crop production techniques. Current discussions on the contribution of biofuels to cover the increasing demand for energy are focusing on the use of abandoned agricultural land rather than on the use of cropland for biofuel production (Bustamante et al., 2009). Field et al. (2008) estimated that globally 4.7–5.8 Mio. km2 of agricultural land are currently abandoned and that biofuels grown on this land may provide 32–41 EJ year−1 or approximately 5% of the expected global energy demand in 2050.

Moreover, it is critical to discuss the effects of large-scale biomass production on other ecosystem services such as biodiversity, water availability, and quality of regional air pollution. A recent study by Baum et al. (2009) came to the conclusion that SRC may even increase biodiversity compared to adjacent arable fields or coniferous forests. With regard to faunal diversity, the increasing cultivation of short-coppice rotation systems may lead to a slight increase in biodiversity in cleared agricultural landscapes (Schulz et al., 2009). For the United States of America Williams et al. (2009) came to the conclusion that increased use of SRC may promote landscape restoration and diversity, thereby enhancing species biodiversity and natural habitats. However, cleared agricultural landscapes are not typical for Romania. A review of biodiversity impacts by SRC introduction in the United Kingdom and Europe also found generally improved floral biodiversity and avian diversity vs. arable crop production but notes that many SRC systems may represent poorer habitats than ancient woodlands, wet meadows or unimproved grassland (Rowe et al., 2009).

Besides biodiversity one also needs to consider the water footprint of large-scale poplar plantations for producing lignocellulosic bioenergy. As can be depicted from Fig. 2, large potential areas for SRC on marginal land have unfavorable conditions with regard to annual rainfall availability. SRC plantations were also found to have larger transpiration losses than permanent grassland or winter wheat in the United Kingdom due to increased rooting depths and thus enhanced water access (see Rowe et al., 2009). Although the production of SRC is likely to have much lower total water demands than the production of corn-grain crops, growing SRC on marginal land will potentially require substantial irrigation to ensure economic viability (Williams et al., 2009). A study evaluating water demand of poplar plantations in Spain vs, their potential for energy production (expressed in savings in CO2 equiv.), came to the conclusion that the consumption of water required to avoid 1 kg of CO2 equiv. is 4.6 m3. In terms of unit of energy obtained this is equivalent to 45 m3 GJ−1 (Sevigne et al., 2011). This implies that water shortage in some areas of Romania, such as for the eastern lowlands, may further increase due to reductions of groundwater recharge for areas with poplar plantations as compared to those planted with crops and the possible need for irrigation to maintain high productivity.

Our study shows that lignocellulosic bioenergy production via short-coppice rotation poplar plantations have a potential to significantly contribute to the national energy supply. However, this will require that substantial areas are utilized for SRC plantations, since expected yields are approximately in a range of 10–13 t ha−1 year−1. Even though at least 10% of total agricultural land in Romania is set-aside at present, a potential target to cover approximately 20% of national energy consumption will require about 2 million ha planted with SRCs. This is twice the amount of current set-aside areas and this scenario would probably interfere with conventional crop production. Hence, expansion of SRCs on such large areas has to be considered critically. In addition to conflicts with crop production needs we also did not explicitly account for SRC impacts on ecosystem services or land-use change effects on the national greenhouse gas balance (Aylott et al., 2010).

Future advances in plant breeding and a specific selection of cultivars may help to increase yields and to reduce land demand. Nevertheless, the important issue of water demand and feedbacks of large-scale SRCs on regional water availability in dryer regions of Romania require special attention.


This research was carried out in the framework of the BioForRisk project funded by the “Daimler Fonds im Stifterverband für die Deutsche Wissenschaft”. Additional funding was received by the Federal Ministry of Education and Research in the framework of the ProBioPa project. C. Werner would like to acknowledge financial support by the research funding program LOEWE “Landes-Offensive zur Entwicklung Wissenschaftlich-ökonomischer Exzellenz” of Hesse's Ministry of Higher Education, Research, and the Arts.