I. Callesen, Biosystems Division, Risø National Laboratory for Sustainable Energy, Technical University of Denmark – DTU, Building 309, PO Box 49, Frederiksborgvej 399, DK-4000 Roskilde, Denmark, tel. +454525 4672, fax+45 4593 3435, e-mail: firstname.lastname@example.org
A further increase in nitrogen (N) intensive biomass supplies to substitute fossil carbon sources implies inclusion of additional reactive nitrogen (Nr) into the biosphere. A Danish model study compared low-intensity managed seminatural beech forest and a winter wheat system with respect to N losses and greenhouse gas (GHG) emissions. Losses of reactive N to air and groundwater per unit of energy produced were four to six times higher for the winter wheat system. The energy efficiency was an order of magnitude higher in the forest system, whereas the related GHG emission reduction by fossil coal substitution differed by <25%. The question is whether a low or a high intensity of cultivation yields the best overall ecosystem service performance? Given the detrimental effect of excess reactive N on natural ecosystems, we suggest that bioenergy production from unfertilized forest with seminatural structure and function should be preferred over N-intensive crop production.
Biomass (from natural and cultivated ecosystems) is globally a limited resource as available land is limited and the inputs, which are needed for cultivation are costly in terms of energy input (EI), e.g. fertilizer (organic or nonorganic), fuel, machinery and human labour. Since biomass will be increasingly important as an energy source in a fossil-free future, it is now relevant to see forestry and agriculture as competing land uses, but also to address the role of land cultivation as a potential driver for climate change.
Nitrogen (N) is the nutrient that generally limits plant growth, and the global cycling of biologically available N has been doubled by humans to reach today's agricultural production yields (Erisman et al., 2008). The elevated N cycling has caused eutrophication, acidification, emissions of nitrous oxide (N2O) [strong greenhouse gas (GHG)] and higher species extinction rates (Vitousek et al., 1997). An increase in fertilized cropland areas on the expense of natural ecosystems will aggravate this trend.
Biomass for bioenergy has traditionally been extracted from the forest either as fuel wood or in secondary use phases as wood waste, but agricultural feedstocks and agricultural waste are increasingly important. Most agricultural crops require short-term, e.g. annual, rotations with frequent disturbance of the soil environment. The production of bioenergy from starch, oil or sugar crops utilizes feedstock resources that could otherwise be used for human or animal nutrition. It is thus relevant to analyse and evaluate some of the characteristics and consequences of biomass utilization from low- and high-intensity production systems. There is a strong focus on limiting the GHG emissions from bioenergy (European Commission Directive, 2009), but the importance of minimizing the use of N fertilizer and thereby the cycling of reactive nitrogen (Nr) needs to be realized (Melillo et al., 2009; Erisman et al., 2010). The question is how land cultivation for bioenergy performs in terms of N use efficiency, energy use efficiency and adverse environmental effects per unit of energy produced, and especially whether a low or a high intensity of cultivation yields the best overall ecosystem service performance of the entire landscape?
Here, we investigate the efficiency of biomass production in Danish (humid temperate) ecosystems by studying the low- and high-end extremes of management or cultivation intensity. We compare biomass utilization from managed seminatural forest and from a winter wheat field. The analysis focuses on the N use efficiency during biomass production, and the emissions of the GHGs N2O, methane (CH4) and carbon dioxide (CO2) created by the systems (Fig. 1). The acceleration of the N cycle as an example of an exceeded planetary boundary (Rockström et al., 2009) is our main focus, and we thus tone down the analyses of carbon (C) balances of land conversion for biomass cropping (e.g. Fargione et al., 2008; Searchinger et al., 2008). The effect of cultivation is evaluated by using the potential natural vegetation type as a reference for comparison.
Materials and methods
Two biomass utilization systems with very low (natural forest) vs. high (wheat) cultivation intensity were compared, using the seminatural potential natural vegetation in a forest reserve (FR) as reference system. The geographic location of the study was Eastern Denmark (∼55°N, 11°E) that holds a humid temperate climate and nutrient-rich sandy loam soils; and a high N load due to intensive agriculture and a high population density. The forest vegetation types were a broadleaved, deciduous beech-dominated FR, Suserup (FR) and a managed beech forest in continuous cyclic management (FM) with natural regeneration (Table 1). The site Suserup Skov (19.3 ha) is one of the rare existing examples of unexploited seminatural FRs in NW Europe (Emborg et al., 2000). The high-intensity cultivation system was a winter wheat (Triticum aestivum L.) field (WW). Only the FR refers to a specific site, whereas the remaining types are averages for growth conditions typical for the region.
Table 1. Ecosystem types and data sources from four vegetation types belonging to eastern Denmark
Forest reserve Suserup Skov, Sorø, Denmark
55°22′N, 11°34′E, 19.2 ha forest reserve dominated by beech (Fagus sylvatica L.)
The studied systems are illustrated in Fig. 1 and all data are scaled to 1 ha land area. We apply a low-intensity forest management regime to demonstrate the minimal effort required to manage forests while maintaining natural structure and functioning as far as possible. In the FM, mature stems for harvesting are selected according to the target dimension principle, whereby trees are harvested when they reach a certain diameter, e.g. 50 cm at breast height (stem diameter 1.3 m aboveground). This is considered a fundamental principle in close-to-nature forest management (Gamborg & Larsen, 2003). The forest maintains natural regeneration by seed dispersal in mast (seed production) years. Application of 4 t of low-reactive wood ash every 40 years corresponding with the equivalent amount of nutrients in biomass removals is carried out to return soil nutrients. The winter wheat receives 166 kg N ha−1 as NPK fertilizer, 1 t lime ha−1 yr−1, i.e. 4 t ha−1 every 4 years, and pesticide treatments and no ash recycling. Machine operations include all field operations, transport of harvest yield to farm gate and drying of grains (Danish Agricultural Advisory Service, 2007).
The biomass was harvested from the managed ecosystems and incinerated at 90% efficiency in a combined heat and power (CHP) plant equipped with a de-NOx facility that strips the NOx of the flue gas, e.g. by using NH3. In both cases immediate incineration of such potentially high value products is economically inefficient, but since we focus on N efficiency, we apply a simple utilization step with equal efficiency in both systems. We leave out the time dimension of the wood first being used for wood products or the possibility of using the wheat for food, feed or liquid biofuel.
Comparisons of N, energy and GHG balances were made. We focused on C storage in aboveground biomass (t C ha−1), GHG emissions from the ecosystems (CO2, CH4 and N2O) as well as upstream emissions from manufacturing of products and services used. Energy yield per year (EYNPP) of aboveground net primary production (NPP) was calculated as tonnes of dry matter (DM) ha−1 yr−1 and converted to GJ ha−1 yr−1 by the lower heating value at 0% water content. The EI required to harvest this yield (GJ ha−1 yr−1) was calculated based on energy use values in Dalgaard et al. (2001), B. Talbot (personal communication) and the database http://www.biolexbase.dk (accessed September 2009). The net energy yield (EYnet) was calculated as the energy content of the harvested biomass (EYharvest) minus the EI. The ratio between net energy yield and EI was calculated as EYnet : EI (Table 2). Energy use for transport of the biomass from the farm gate to the power plant and the power plant facility itself was not included.
Table 2. Energy inputs and outputs of two pairs of Danish seminatural and managed temperate ecosystems
Forest reserve (FR)
Managed forest (FM)
Winter wheat (WW)
Lower heating values adapted from http://www.biolex.dk. Grass heating value assumed similar to straw ∼18.2 MJ kg−1 dm.
Aboveground NPP calculated as EYharvest+10% of straw biomass to include stubble.
For the forest ecosystems, aboveground C stocks were live and dead biomass under winter conditions. In Suserup Skov, the C pool was estimated from a standing biomass volume of 670 m3 ha−1 that included live roots with a diameter over 3 cm (Vesterdal & Christensen, 2007). In the calculation, the DM content was assumed to be 0.56 kg dm−3 and the C content was 0.5 kg C kg−1 dry biomass. Also dead biomass was assessed (Table 1). For the FM, the average stocking was set to 370 m3 ha−1 based on forest statistics (Nord-Larsen et al., 2008) and converted to C by the same factors. A dead wood fraction of 1.8% and a belowground live root biomass of 17% of aboveground live biomass was estimated based on distribution data for broadleaf forests in Denmark (Nord-Larsen et al., 2008). Forest soil C to 100 cm depth was either measured on site as an average of four soil pits (Vesterdal & Christensen, 2007) or based on empirical data for fertile loamy soils (alfisols) situated in forest (Vejre et al., 2003).
Winter wheat aboveground biomass C was taken as the autumn post harvest level. The C-tool model (Petersen et al., 2002) was used to estimate the soil C pool of winter wheat on sandy loam soils with continuous complete straw removal.
Inclusion of reactive N due to cultivation
N flows resulting from the cultivation such as energy use-related NOx emission, inputs (ambient N deposition and fertilizer) and outputs (harvest, leaching and denitrification) was established based on literature values in order to assess the inclusion of reactive N and direct losses of N to water and atmosphere as nitrate and N2O following cultivation. Biological fixation of N2 and loss of N2 due to denitrifrication was ignored. Atmospheric N inputs were set to 10 kg N ha−1 yr−1 for low vegetation and 20 kg N ha−1 yr−1 for forest due to a higher vegetation roughness (Gundersen, 2008).
The global warming potential in terms of CO2 equivalents (Forster et al., 2007) included CO2 (GWP 1), CH4 (GWP 25) and N2O (GWP 298) from 1 ha of land.
For CO2 exchange, we used a budget technique that was based on both measured and statistical data for aboveground NPP. The annual net exchange of CO2 was assumed to be neutral in the seminatural ecosystems (+/−interannual climate-related variation), which means that production equalled decay rates or, in other words, that CO2 uptake due to photosynthesis equalled emission due to respiration in the FR. For the two managed systems, we assume that CO2 harvested is re-emitted immediately due to decay or combustion, whereas decay of aboveground harvest residues (small branches and twigs; straw stubbles) are neglected. In the FM only 67% of annual increment is harvested and the rest accumulates in live woody biomass. The expansion in standing wood volume is possible, since the FM type had less standing biomass than the FR and the increase in wood volume may thus continue for decades. The live biomass in the FM increases by 2.6 m3 ha−1 yr−1 equal to 2.7 t CO2-eq ha−1 yr−1 that may also be harvested for bioenergy when the standing wood volume reaches an adequate level. The annual production of fine roots, twigs and small branches is assumed to balance out with the decay of the same pools. The wood decay of live larger branches and stems due to, e.g., wind throw, herbivory and wood destructing fungi was set to a small fraction of 0.1% yr−1. The net CO2 balance of the soil is assumed neutral.
The N2O emissions related to the cycling of reactive N in ecosystems, and emissions relating to fossil energy used in cultivation were assessed using literature values. The references are given in Table 4. On site N2O emissions from added N fertilizer were calculated as 1% of N fertilizer additions (IPCC, 2006). CH4 fluxes were derived from reported measurement data of a mature beech forest situated 15 km NW of the Suserup forest (Skiba et al., 2009).
Table 4. Greenhouse gas accounts based on literature values of similar ecosystems and own conservative assumptions
Assumption: CO2 taken up is reemitted immediately due to decay or incineration. In managed forest, only 67% of annual increment is harvested. Decay of aboveground harvest residues (small branches and twigs; straw stubbles) neglected in this calculation. Live woody biomass decays with 0.1% yr−1.
Amount of ash corresponding with harvest removals (Møller, 2001).
Off site emissions are related to upstream or downstream processes, e.g. CO2 emissions emanating from products and services used for cultivation such as fertilizer, lime, diesel for field operations and drying, machine manufacture and pesticide production. These emissions were included in the GHG budget. A fraction of 2.5% of nitrate leaching was attributed to N2O emissions off site (Illerup et al., 2006). Up- and downstream emissions related to the manufacture and operation of a CHP plant were assumed to be similar for the two managed systems and thus omitted.
Other GHG exchanges pertain to CH4 consumption by methanotrophs. The GHG balance further includes emissions due to ash application and liming (44 g CO2 100 g−1 CaCO3 assuming full release in the year of application).
Efficiency metrics for land use, fertilizer use and GHG reduction potential were defined. To measure land-use efficiency, the amount of energy harvested was expressed as energy yield per unit land converting by CHP and deducting the input energy [M1: EYnet(conv.); GJ ha−1 yr−1], and the net output : input ratio [M2: (EYharvest–EI)/EI; unitless]. The efficiency of using reactive N was defined as the amount of converted bioenergy produced per kg of N used as fertilizer+N deposition [M3: Nr : EYnet(conv); kg Nr GJ−1]. The N deposition as a total of NHy and NOx immissions was loaded on the N budgets since it may reach significant amounts. Especially the forest may act as a filter for air pollution, which would be distributed differently in the environment if the forest was not present. GHG efficiency was expressed as the net GHG emission per unit of final energy [M4: GWP:EYnet(conv.); kg CO2-Eq GJ−1]. The avoided fossil CO2-equivalents (M5: Avoided GWP; t CO2-Eq ha−1 yr−1) was the emission from the amount of coal that could be replaced by the bioenergy produced in each of the two management types with an emission factor of 95.3 kg CO2-Eq GJ−1 substituted coal (Danish Energy Agency, 2008). When combining M4 and M5, the overall emission reduction of replacing coal with biomass from the two management types forest and winter wheat was obtained (M6: GHG reduction; t CO2-Eq ha−1 yr−1).
The C storage in the forest reserve indicates that current C stores in managed forest and winter wheat are diminished due to C losses by ancient deforestation and cultivation of the land (Fig. 2). In the FR, 382 t C ha−1 was stored in live and dead wood, forest floor and mineral soil. Across the three ecosystem types, soil organic C (SOC) stores were in the range 88 to 132 t C ha−1. The highly variable SOC stores on these upland soils reflect that e.g. drainage regime and land-use management history may influence SOC storage. The FR demonstrates that SOC storage may be high under undisturbed conditions. SOC to 100 cm depth, including forest floor, constituted up to about 40% of the total C storage in the two forest types (above- and belowground), Fig. 2, possibly indicating a SOC storage potential when FM is changed to FRs (Vesterdal & Christensen, 2007). The aboveground C storage was close to nil in the winter wheat system (Fig. 2). C storage in trees provides flexibility in the biomass utilization: NPP can be stored in living biomass to be utilized at a later stage or be a permanent reservoir (Kirschbaum, 2003; Righelato & Spracklen, 2007).
Efficiency of N use and energy use
In comparison with the FR, the production (aboveground NPP) in the FM was higher as trees are younger and more vital. The NPP in DM aboveground biomass was 3.5 t DM ha−1 yr−1 in the FR, Table 2. In the FM, the amount of harvested biomass was 3.5 t DM ha−1 yr−1 (out of a total NPP of 5.3 t DM ha−1 yr−1 in the beech forest). The EI for managing the beech forest was 0.6 GJ ha−1 yr−1 for harvesting and transport of stems and the harvested energy EYharvest was 68 GJ ha−1 yr−1. Biomass for bioenergy could be obtained with an net energy output : input ratio, EYnet : EI, of 122 : 1 (Table 2).
Fertilizer application to winter wheat increased the biomass production to 12.4 t DM ha−1 yr−1, but also gave rise to N losses with seepage water as well as direct and indirect N2O emissions (Table 3). In the winter wheat system, the harvested energy yield was 174 GJ ha−1 yr−1 and the EI for cultivation and harvest was 18 GJ ha−1 yr−1 giving rise to an EYnet : EI of 9 : 1. The main EIs came from NPK fertilizer (Table 2).
Table 3. Annual inputs and outputs of nitrogen (N; kg N ha−1 yr−1)
Forest reserve (FR)
Managed forest (FM)
Winter wheat (WW)
Assumption: No N2O emissions occur during N fertiliser production at the fertiliser plant. Harvested N is captured in a de-NOx facility after combustion in the CHP plant and transformed to inert N2. Biological fixation and loss of N2 due to complete denitrification in soil is assumed to be small and is thus ignored.
N flows on site with significance for the green house gas budget were assessed (Table 3). Direct N losses were 5 kg N ha−1 yr−1 for forest and 42 kg N ha−1 yr−1 for the winter wheat, mainly due to nitrate leaching. The N not accounted for indicated accumulation of N in the ecosystems in the range 4–15 kg N ha−1 yr−1. However, sources of uncertainty include the import and export of inert N2 by fixation and complete denitrification.
Apart from upstream sources the main sources of GHG emissions are denitrification both occurring on site and following nitrate leaching. The overall GHG balance for the unmanaged FR was almost neutral, ∼0.2 t CO2-Eq ha−1 yr−1, whereas the FM accumulated 2.7 t CO2-Eq ha−1 yr−1 due to an increase in aboveground biomass. As illustrated in Fig. 2, accumulation of live aboveground biomass could take place in the FM. The cultivation and harvest of 1 ha of winter wheat emitted 3 t CO2-Eq ha−1 yr−1 and energy use and N2O emissions contributed equally to the net emission estimate (Table 4).
N introduced with fertilizer or air pollution and removed with the harvest will cascade through the product chain and eventually end up in the environment causing terrestrial and aquatic eutrophication and further N2O losses. Figures 3 and 4 illustrate efficiency metrics for loss of reactive N (M3, Fig. 3), land use and energy use of bioenergy production (Fig. 4, M1 and M4–M6). The N efficiency denotes the amount of Nr inclusion per GJ energy yield. Harvest of biomass in unfertilized low-intensity forest management as a utilization strategy for bioenergy (and other nonfood) production is much more N efficient than high-intensity winter wheat. In the winter wheat field, the Nr : EYnet(conv) value was 0.3 kg Nr GJ−1and in the forest the value was 0.1 kg Nr GJ−1. Furthermore, in the forest case the Nr represents N that is already included in the biosphere e.g. due to combustion processes outside the system, whereas the Nr in the winter wheat system is newly included Nr from synthetic fertilizer. The N cascade through ecosystems stops with the harvested biomass that is incinerated in a CPH plant with a de-NOx facility. If the de-NOx equipment was not available the Nr included in the biosphere from wheat would be 1.2 kg Nr GJ−1 and in the FM it would be 0.2 kg Nr GJ−1.
The gross energy yield per hectare is high in the wheat field and relatively low in the FM. The GHG balance was −45 kg CO2-Eq GJ−1 for managed beech forest (a net uptake from the atmosphere) and 22 kg CO2-Eq GJ−1 for winter wheat (a net emission to the atmosphere). If these feedstocks replace coal for incineration in a heat plant, the avoided emission from coal of would be 6 t CO2-Eq ha−1 for FM and 15 t CO2-Eq ha−1 for wheat grain and straw (M5). If the GHG net balance from cultivation is added (−3 and +2.7 t CO2-Eq ha−1, Table 4) the avoided emission by replacing coal are 9 t CO2-Eq ha−1 for FM and 12 t CO2 Eq ha−1 for winter wheat (M6).
Inherent soil productivity
These examples are valid for productive, fertile soils in eastern Denmark. Winter wheat on poor, sandy soils in western Denmark will produce about 65% of the yields on loamy soils presented here (Danish Agricultural Advisory Service, 2007). However, irrigation in dry years may be necessary to achieve these lower yields on the sandy soils. Nemoral forest species such as beech and oak are less productive on coarse sandy soils and may produce only 20–30% of the NPP yielded on sandy loam soils (Callesen et al., 2006). The relative difference in NPP between forest and winter wheat will thus be larger on sandy soils reflecting the inherent soil productivity. In our example on sandy loam soils, the energy productivity per land area is also clearly lower in the forest. However, an evaluation of the avoided greenhouse gas emissions from forest biomass viz. winter wheat biomass from one hectare of land turns out to be more equal (Fig. 4, M6).
Based on literature data, two ecosystem types of very low viz. high cultivation intensity were used to demonstrate the efficiency of N-intensive land cultivation. Inclusion of reactive N into the biosphere either by synthetic or biological fixation is a prerequisite for plant growth. When the purpose is climate friendly energy production (or other nonfood purposes) rather than food production, requirements for GHG efficiency and limited adverse environmental impacts should be very strict, because the reference system is potential natural vegetation (lignocellulosic biomass) and not other food production (plant-based starches, oils and sugars). Although the winter wheat produces more NPP per area, the consequence is loss of potential natural vegetation (which in this case happened during the medieval period) and four to six times more reactive N in the biosphere per GJ produced. We calculated that cultivation increased the NPP in both FM and winter wheat in comparison with the FR. Examples of energy output : input ranges in FM are 30 : 1 to 20 : 1 (including chipping and 50 km transport, Hakkila & Parikka, 2002), whereas the selected case represents the most ideal conditions for close-to-nature forestry. Most FMs today are not yet in a state that allows this type of forestry; a transition of the silviculture towards close-to-nature forestry is needed. However, even a 10 times lower energy output : input ratio for the FM would increase the Nr : EYnet(conv) for FM by <25%. If the emission factor for N2O from fertilizer is raised to 5% of additional Nr (see e.g. Crutzen et al., 2007) rather than 1%, the total GHG emission reduction (M6) is only 8 t CO2-Eq ha−1 yr−1 for wheat, but still 9 t CO2-Eq ha−1 yr−1 for the FM. The key issue is the use of N fertilizer since it manyfold exceeds the increase in Nr caused by natural fixation of N in the potential natural vegetation.
Efficiency of reactive N inclusion in the biosphere
The analysis indicates that high-intensive N-aided (either synthetically or biologically fixed N) agricultural production should be aimed at the necessary food production. The use of agricultural residues for energy production (e.g. second-generation bioethanol) and organic N-fertilizer may reduce the net green house gas emissions induced by the cultivation of land for food, but only if the reactive N in the agricultural waste replaces synthetically fixed N elsewhere. The question remains, how to optimize the ecosystem services per unit of reactive N – by intensifying low-intensity systems or extensifying high-intensity cultivation systems (Fig. 1)? Between the two extremes that were analysed here, a domain of cultivated ecosystems exists where bioenergy, and also food may be produced. If perennial bioenergy crops with high C : N ratio (lignocellulosic crops) are to be fertilized, the overall GHG balance of fertilization accounting for the full effect of cascading N should indicate an additional GHG sink in comparison with the unfertilized case. In areas with low N deposition increasing N limitation on biomass production due to biomass harvest is also a concern. A careful allocation of land area for protected nature reserves with potential natural vegetation, low- and high-intensity cultivation systems may provide the best obtainable provision of ecosystem services at the landscape scale.
It is possible to produce biomass for energy from ecosystems that do not require harmful Nr inclusion into the biosphere and that closely resemble species composition, structure and function of the potential natural vegetation. However, removal of NOx emissions from combustion processes (cf. Fig. 3) is important in any case to secure the least possible negative impact of reactive N on natural ecosystems following biomass utilization.
The Villum Kann-Rasmussen foundation financed post doc studies for Ingeborg Callesen (Grant number. VKR09b-015).