Modelling greenhouse gas emissions from European conventional and organic dairy farms

Modelling greenhouse gas emissions from European conventional and organic dairy farms

Agriculture, Ecosystems and Environment 112 (2006) 207–220 www.elsevier.com/locate/agee Modelling greenhouse gas emissions from European conventional...

414KB Sizes 0 Downloads 86 Views

Agriculture, Ecosystems and Environment 112 (2006) 207–220 www.elsevier.com/locate/agee

Modelling greenhouse gas emissions from European conventional and organic dairy farms J.E. Olesen a,*, K. Schelde a, A. Weiske b, M.R. Weisbjerg c, W.A.H. Asman a, J. Djurhuus a a

Department of Agroecology, Danish Institute of Agricultural Sciences, P.O. Box 50, DK-8830 Tjele, Denmark b Institute for Energy and Environment, Torgauer Strasse 116, D-04347 Leipzig, Germany c Department of Animal Nutrition and Physiology, Danish Institute of Agricultural Sciences, P.O. Box 50, DK-8830 Tjele, Denmark Available online 12 October 2005

Abstract Agriculture is an important contributor to global emissions of greenhouse gases (GHG), in particular for methane (CH4) and nitrous oxide (N2O). Emissions from farms with a stock of ruminant animals are particularly high due to CH4 emissions from enteric fermentation and manure handling, and due to the intensive nitrogen (N) cycle on such farms leading to direct and indirect N2O emissions. The whole-farm model, FarmGHG, was designed to quantify the flows of carbon (C) and nitrogen (N) on dairy farms. The aim of the model was to allow quantification of effects of management practices and mitigation options on GHG emissions. The model provides assessments of emissions from both the production unit and the pre-chains. However, the model does not quantify changes in soil C storage. Model dairy farms were defined within five European agro-ecological zones for both organic and conventional systems. The model farms were all defined to have the same utilised agricultural area (50 ha). Cows on conventional and organic model farms were defined to achieve the same milk yield, so the basic difference between conventional and organic farms was expressed in the livestock density. The organic farms were defined to be 100% self-sufficient with respect to feed. The conventional farms, on the other hand, import concentrates as supplementary feed and their livestock density was defined to be 75% higher than the organic farm density. Regional differences between farms were expressed in the milk yield, the crop rotations, and the cow housing system and manure management method most common to each region. The model results showed that the emissions at farm level could be related to either the farm N surplus or the farm N efficiency. The farm N surplus appeared to be a good proxy for GHG emissions per unit of land area. The GHG emissions increased from 3.0 Mg CO2-eq ha1 year1 at a N surplus of 56 kg N ha1 year1 to 15.9 Mg CO2-eq ha1 year1 at a N surplus of 319 kg N ha1 year1. The farm N surplus can relatively easily be determined on practical farms from the farm records of imports and exports and the composition of the crop rotation. The GHG emissions per product unit (milk or metabolic energy) were quite closely related to the farm N efficiency, and a doubling of the N efficiency from 12.5 to 25% reduced the emissions per product unit by ca. 50%. The farm N efficiency may therefore be used as a proxy for comparing the efficiencies of farms with respect to supplying products with a low GHG emission. # 2005 Elsevier B.V. All rights reserved. Keywords: Farm model; Carbon dioxide; CO2; Methane; CH4; Nitrous oxide; N2O; Conventional farming; Organic farming; Dairy farming

1. Introduction The emissions of carbon dioxide (CO2), methane (CH4) and nitrous oxide (N2O) from agriculture together account * Corresponding author. Tel.: +45 89991659; fax: +45 89991619. E-mail address: [email protected] (J.E. Olesen). 0167-8809/$ – see front matter # 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.agee.2005.08.022

for approximately one-fifth of the annual increase in radiative forcing leading to climate change (Houghton et al., 2001). This proportion is even higher, if indirect energy use by agriculture in fertilisers, machinery, etc. are included in the estimates. Agriculture accounts for 50% of the anthropogenic emissions of CH4 (Mosier et al., 1998a). The major

208

J.E. Olesen et al. / Agriculture, Ecosystems and Environment 112 (2006) 207–220

agricultural sources of CH4 are enteric fermentation in ruminant animals and anaerobic turnover in rice paddies, whereas anaerobic fermentation in animal manures accounts for 8% of the agricultural CH4 emissions. Agriculture accounts for 80% of global anthropogenic N2O emissions, and about 40% of this originates from manure management (Mosier et al., 1998b). However, the uncertainties in the emissions are large and so is the variation in time and space of fluxes from different sources. This makes the identification and implementation of effective measures to reduce emissions from agriculture difficult (Oenema et al., 1998; Velthof et al., 1998). Emissions from farms with a stock of ruminant animals are particularly high due to CH4 emissions from enteric fermentation and manure handling, and due to the often intensive nitrogen (N) cycling on such farms leading to both direct N2O emissions from the soils and indirect N2O emissions from N lost by ammonia (NH3) volatilisation and nitrate leaching (Oenema et al., 1998; Monteny et al., 2001). A proper account of all emissions requires an integrated model of all C and N flows on the farm. Such a model can then further serve as a basis for evaluating mitigation options to reduce the emissions. This paper presents a model for estimating greenhouse gas (GHG) emissions from dairy farms in Europe. The model was constructed to allow farm level mitigation options to be analysed (Weiske et al., 2005). The model considers emissions of CH4 and N2O from the farm management, and emissions of CO2, CH4 and N2O from imported energy, feedstuffs, fertilisers, etc. For simplicity the model mostly applies an emission factor approach for the estimation of N2O emissions, whereas more complex algorithms are used to estimate CH4 emissions from enteric fermentation and manure storage. Changes in soil carbon (C) storage and energy used for production of machinery and buildings are not considered in the model. The aims of this paper are three-fold: (i) to present the model, (ii) to analyse the effect of organic and conventional management practices on whole farm GHG emissions from dairy farms and (iii) to compare emissions from dairy farms in different regions of Europe.

2. Materials and methods 2.1. FarmGHG model FarmGHG is a model of C and N flows on dairy farms. The model was designed to allow quantification of all direct and indirect gaseous emissions from dairy farms, so that it can be used for assessment of mitigation measures and strategies. The model includes all GHG emissions on the farm, including indirect N2O emissions associated with N losses, and pre-chain emissions from imports of products, but not emissions after the exported products have left the farm. The imports, exports and flows of all products through the internal chains on the farm are modelled (Fig. 1). The model thus allows assessments of emissions from the production unit and all pre-chains. The model includes the N balance, and it allows calculation of environmental effect balances for GHG emissions (CO2, CH4 and N2O) and eutrophication (nitrate and NH3). The model is described in detail by Olesen et al. (2004). The model includes not only the farm gate budget components (input/output), but also the internal flows in the system (Watson and Atkinson, 1999). These internal flows are represented as flows between compartments in the farm system. The model also explicitly includes all C and N losses except for soil respiration and N2 emissions from soils. The energy use is calculated for each compartment and is converted to pre-chain emissions of CO2 and other greenhouse gases. Milk production and herd size were given from the definition of the model farms. The imports of C and N in fertiliser, bedding, feed, seed and irrigation resulted from the desired milk production, and from the specification of model farms, but were distributed between the model compartments in the simulations. This resulted in an export of C and N in milk and meat. The farms were assumed to be operated by best management, and not to export manure. Net crop production that was not used for feed was either exported or added to the farm manure storage. The model flows are in principle updated using a time step of 1 month. However, in order to properly reflect the

Fig. 1. Flows of C and N in and out of the total model farm system and between compartments within the system represented in FarmGHG.

J.E. Olesen et al. / Agriculture, Ecosystems and Environment 112 (2006) 207–220

209

Table 1 Order of steps used for updating the model state Step

Procedure

1

The demand of the crops for nitrogen from manure is estimated and satisfied if manure is available from the manure storages, i.e. manure is moved from the manure storages to the respective fields. Mineral nitrogen is added on conventional farms to achieve the required N demand in the specific month The animal compartments are updated and milk and meat is exported from the farm. Emissions of methane from the animals are estimated Feed is transferred from the feed storage to the animals to satisfy the need of the animals. Fresh feed from grazing, etc. are also virtually transferred to the animals via the feed storage The produced urine and faeces is allocated to grazed fields and to the house in proportion to the time spent in house or in the fields Bedding and water is added to the house according to rates needed for the specific house and animal numbers. The need for bedding and water is reduced in proportion to the time spent in the house The house compartment is updated and emissions of methane and nitrous oxide are estimated Manure is transferred from the house to the manure storages, and a methane production capacity of the organic matter is assigned to each manure type The manure storage is updated and emissions of methane and nitrous oxide are estimated. If the manure storage includes a biogas digester then additional estimates of electricity and heat production are made The crop rotations are updated and emissions of methane and nitrous oxide are estimated

2 3 4 5 6 7 8 9

flows and emissions, daily time steps are used to simulate the flows between the animals, house and manure storage compartments. The model is initialised by clearing the contents of feed and manure storages on the farm. However, this gives an unrealistic representation of the manure storages, and the model was therefore run for an initial year to equalise compartments before doing the actual simulations. The steps used for each monthly update of the model are shown in Table 1. The balance of the feed store is adjusted at the end of the year by importing any deficits compared with what has been used on the farm for feeding. If there are surpluses in one type of forage crop, then this will be used to reduce deficits of other types of forage crops. Any deficits of forage crops will be imported as silage. Surplus of grass from grazed fields is returned to the fields as residues, and surplus of other types of forage crops will be added to the manure store. All other plant products are sold from the farm. The feed plans for the farms should be adapted to minimise the feed surplus that is added to the manure store. The model allowed different methodologies for emissions estimations to be used. The tier 1 and tier 2 methodologies of the IPCC (1997) and the IPCC Good Practice Guidance (IPCC, 2000) were implemented. In addition, a default FarmGHG methodology was used to estimate flows and emissions of CH4 and N2O. The estimated emissions of CO2, CH4 and N2O were compared in terms of their 100-year global warming potentials (CO2-equivalents), which on a weight basis relative to CO2 was set to a factor of 21 for CH4 and 310 for N2O. 2.1.1. Animals The model distinguishes two groups of livestock on the farm, cows and young stock (heifers). The feed is defined by a feed plan, which specifies the relative proportions of feed

used on a dry matter basis. There are separate feed plans for cows and heifers, and the feed plans can differ between summer and winter. During the summer period, cows and heifers will usually be grazing or given fresh feed. The feed energy requirement is specified in Scandinavian Feed Units (FU). Scandinavian FU is a net energy (NE) evaluation system, and one FU is equivalent to 7.89 MJ NE (Weisbjerg and Hvelplund, 1993). The energy requirements of the dairy cow vary during the lactation period (time since last calving) due to varying milk yield and foetal growth. Nevertheless, LR (1999) estimated an annual feed requirement of 6100 FU for a dairy cow weighing 600 kg and yielding 7800 kg energy corrected milk (ECM). This value has to be scaled to the actual milk yield for the cows in question. At 100% feed efficiency, producing 2.5 kg ECM requires an energy intake of 1 FU (LR, 1999). At a feed efficiency of 83%, common at Danish dairy farms (Poulsen et al., 2001), production of 2.075 kg ECM requires 1 FU. The equation used for downscaling is thus FUdown ¼ FU0 

ECM0  ECMdown 2:075

(1)

where FUdown is the actual feed demand (FU), ECMdown the actual annual milk yield (kg ECM) and the benchmark values FU0 and ECM0 are the 6100 FU and 7800 kg ECM, respectively. The CH4 emission from enteric fermentation is, in the default FarmGHG methodology, calculated using the empirical equation of Kirchgessner et al. (1995), which is based on the feed intake and the nutrient composition of the feed ration: Methane ¼ a þ 79 CF þ 10 NFE þ 26 CP  212 Fat 1

(2)

where a is the intercept (63 g CH4 day for cows and 16 g CH4 day1 for heifers), CF the intake of crude fibres (kg day1), NFE the intake of nitrogen free extracts

210

J.E. Olesen et al. / Agriculture, Ecosystems and Environment 112 (2006) 207–220

(kg day1), CP the intake of crude protein (kg day1) and Fat is the intake of crude fat (kg day1). 2.1.2. Housing Three different housing types are considered in the model: Tied stall: The animals are tied and fixed within separate stalls that serve as place for both resting and eating. The manure may be handled as either separate (solid and liquid) or slurry. Cubicles: This is a loose housing system, where the animals are allowed to move freely, and where the resting area is separated into separate cubicles. The walkways serve as traffic, manure and exercise areas. The manure may be handled as either separate (solid and liquid) or slurry. Deep litter: This is a loose housing system, where the animals are allowed to move freely. The deep litter mat is also used as resting area. The entire floor area is assumed to be deep litter. There are three types of manure handling systems included in the model: Separate: The urine and faeces is separated in the house in liquid and solid fractions. It is assumed that all of the urine is transferred to the liquid fraction and the faeces and litter is transferred to the solid fraction. In practice some mixing of these fractions may occur. Slurry: The urine, faeces and any straw/litter is mixed and typically collected in a channel or a pit beneath slats. Deep litter: The floor is covered by a layer of straw/litter, and additional straw is added on the top every day. The urine and faeces are collected in the litter, and the deep litter mat builds up over time in the house. In addition, the temperature of the house must be specified, as well as any methods used to clean the floor, as this will affect NH3 emissions. In the IPCC methodology, there is no NH3 emission specifically from animal housing. The NH3 volatilisation in the default FarmGHG methodology is based on the standard values reported by Poulsen et al. (2001). The emission is estimated as a proportion of total-N excreted by the animals. In the default FarmGHG methodology, CH4 emissions from slurry based systems are calculated according to the following equation: ECH4 ¼ km fT VS Bo 0:67

(3)

where VS is the amount of volatile solids or organic matter coming from faeces and bedding material (kg), Bo the maximum CH4 producing capacity (m3 kg1 VS), km a CH4 conversion factor at the reference temperature (15 8C) (proportion of CH4 producing capacity used per

day) and f T is a function of temperature. The standard value used for km is 0.005. The Arrhenius equation is used to describe the temperature dependence of the CH4 emission rate:    DE 1 1  fT ¼ exp (4) R T Tref where DE is the enthalpy of formation (J mol1) taken to be an arbitrary value of 1.22  105 J mol1, R the gas constant (8.31 J mol1 K1), T the actual absolute temperature (K) and Tref is the reference temperature (15 8C or 288.15 K). The available literature does not support any estimation of CH4 emissions from deep litter, solid manure and urine in the house, and such emissions are therefore not currently included in the default FarmGHG methodology. The default FarmGHG methodology includes N2O emissions from slurry, solid (farmyard) manure and deep litter systems. In all cases the parameterisation was taken from Amon (1999). Nitrous oxide emissions from slurry stored in the house are estimated as: EN2 O ¼ 3:33 þ 0:282 max ð15; TÞ

(5)

where EN2 O is the N2O emission (g N2O kg1 N), and T is the temperature (8C). The N2O emission (g N2O kg1 N) from solid manure and deep litter in the house is estimated as: EN2 O ¼ 0:109 þ 0:124 T

(6)

2.1.3. Manure storage Four different manure storage types are considered: Slurry: A mixture of urine, faeces and other organic wastes is stored in a tank. The slurry may be pre-treated by anaerobic digestion. The tank can be covered with a solid cover, straw or with a natural surface crust. Digested slurry will not form a natural surface crust. Liquid manure: The liquid manure is primarily the excreted urine plus some additional water. It is stored in a tank, which can be covered with a solid cover or straw. It will not form a natural surface crust. Solid manure: Solid manure is the faeces and added straw from a separate manure handling system. It is stored in a heap and may compost, if the straw content is high enough, and if it is turned. Deep litter: Deep litter comes from housing systems with deep litter and will have a high straw content, which will make it compost during heap storage. For storage of solid manure and deep litter, the model assumes that no seepage occurs. In practice there is seepage, but under good management this will be collected in the liquid manure store.

J.E. Olesen et al. / Agriculture, Ecosystems and Environment 112 (2006) 207–220

Storage of the manure will lead to loss of both dry matter and N. A major part of the N loss is associated with NH3 volatilisation. For solid manure and deep litter, an additional loss through denitrification is assumed, which mainly results in emission of N2. The emission of N2O is estimated as described below. For estimation of CH4 emissions from slurry tanks in the default FarmGHG methodology, Eqs. (3) and (4) are used. The applicability of these equations is based on studies of CH4 emission from open tanks with surface crust (Husted, 1994; Sommer et al., 2000). The emission rates are reduced by 5% in the model, if a cover is present. This factor was derived from the winter and summer storage experiments of Amon et al. (2004). Methane emissions from solid manures are assumed to decline exponentially with storage time: ECH4 ¼ E0 exp ðatÞ

(7)

is the CH4 emission where ECH4 (kg CH4 Mg1 manure day1), E0 the base emission rate, a a rate parameter and t is the storage time. This equation is used under the assumption that the manure is stored for a period of t = 140 days, and the cumulated emission is calculated as: ECH4 ¼

E0 ½1  exp ðatÞ a

(8)

where ECH4 is the cumulated CH4 emission (kg CH4 Mg1 manure). The CH4 emission from non-composting manure is further related to the ambient temperature, and the base emission rate is therefore calculated as: E0 ¼ Eref fT

(9)

where Eref is the reference emission rate at 15 8C (kg CH4 Mg1 manure day1) and f T is the temperature response function defined in Eq. (4). Eref is set to 0.013 kg Mg1 manure day1, DE is set to 9.0  104 J mol1 and a is set to 0.0347 day1 (Amon, 1999). The CH4 emission from composting manure is determined by the temperature inside the heap, which is hardly influenced by the outside temperature. The parameterisation is based on the work of Hu¨ther (1999) and, to some extent, on the work of Amon (1999). The base emission rate is determined as: E0 ¼ aDM þ bDM DM

(10)

where DM is the manure dry matter content (%), and aDM and bDM are constants. For DM less than 20%, aDM = 0.021 and bDM = 0.423, while for DM larger than 20%, aDM = 0.001 and bDM = 0.025. The parameter a is set to 0.0462 day1. The accumulated CH4 emission from storages with deep litter was assumed to be 0.01% of C stored in the deep litter heap, which is calculated from C in the faeces and applied litter (Sommer, 2001).

211

For slurry stores, the N2O emission rate was assumed to depend on tank surface area. The emission rate is taken as 0.8 g N2O m2 day1 (Amon et al., 2004). For non-composting manure, the N2O emissions are estimated as a function of storage time (Amon, 1999): EN2 O ¼ 1:97 exp ð0:347ðt  8ÞÞ

(11)

where EN2 O is the N2O emission (kg N2O Mg1 manure day1) and t is storage duration (days). Emissions before day 8 are set to 0, because no emissions were measured before this day in the experiment by Amon (1999). For composting manure, the following equation is used: EN2 O ¼ 1:44 exp ð0:347 ðt  30ÞÞ

(12)

where EN2 O is the N2O emission (kg N2O Mg1 manure day1) and t is storage duration (days). The emission before day 30 is set to 0. For both composting and noncomposting manure a total storage time of 140 days was assumed. For deep litter, a N2O emission factor of 0.001 of total N is used (Sommer, 2001). 2.1.4. Fields and crops A range of crop types can be used in FarmGHG; forage crops, cereals and pulses for grain and whole-crop silage, and cash crops. The forage crops include grass-clover (Lolium perenne L. and Trifolium repens L.), red clover (Trifolium pratense L.), alfalfa (Medicago sativa L.) and maize (Zea mays L.). The cereal crops include barley (Hordeum vulgare L.), wheat (Triticum aestivum L.), oats (Avena sativa L.) and triticale (Triticosecale Wittm.), the pulses include pea (Pisum sativum L.) and field bean (Vicia faba L.), and the cash crops include potato (Solanum tuberosum L.). The gross yield of the crops must be specified. For forage crops (e.g. grass) that are harvested more than once, the yield is distributed equally among the months of harvest operations. For forage crops, the yield specified is a gross yield, which is reduced depending on the utilisation of the crop. The harvest loss varies from 10 to 20% depending on crop use, with the highest losses for grazing. The harvest loss is returned to soil as crop residues. The N demand must be specified for each crop and field, and this demand does not include contributions from biological N fixation (BNF). This N demand can be fulfilled by either on-farm available manure, or by purchased mineral N fertiliser for conventional farms. The fertiliser or manure may be split between different application dates as given by the crop specification. The applied N is split equally in the case of multiple application dates. The model will attempt to use all available manure before fertiliser is imported. The amount of manure applied is given by the fertiliser replacement value (FRV), which is defined as the amount of mineral fertiliser that should be applied to replace total N in manure. FRV depends on manure type,

212

J.E. Olesen et al. / Agriculture, Ecosystems and Environment 112 (2006) 207–220

crop type, application method and time of application. The standards presented by LR (2002) have been used as a basis for defining FRV in FarmGHG. In the IPCC (1997) methodology, BNF was only estimated for pulses, whereas BNF from grassland legumes has been included in the methodology of IPCC (2000). In the default FarmGHG methodology, BNF was estimated according to the method described by Høgh-Jensen et al. (2004), which estimates N fixation from all N fixing crops. The default FarmGHG methodology calculates the N in residues (Nres) returned to the field from seeds, harvest losses, grain crops, cover crops and forage crops: Nres ¼ Nseed þ Nloss þ cs uY  cs Ys þ cr Yr

(13)

where Nseed is the N in seeds used for seeding (kg N ha1), Nloss the N in harvest losses for forage crops (kg N ha1), Y the grain yield of grain crops, Ys the straw yield of grain crops, Yr the amount of residues for cover crops and forage crops, cs and cr the N concentrations in straw and cover crops, respectively, and u is the ratio of dry matter in crop residues to crop dry matter yield. In the IPCC methodology, NH3 volatilisation is estimated as a fraction of N in applied fertiliser and manure, and no specific NH3 emissions from housing or manure storages are considered (IPCC, 1997). In the default FarmGHG methodology, the NH3 volatilisation for applied slurry, liquid manure and anaerobically digested manure is calculated using the ALFAM model (Søgaard et al., 2002), which incorporates effects of climatic conditions. The emission is further modified by crop type, crop size and method of application. For farmyard manure and deep litter, simple emission factors are used for ammonia volatilisation after application (Hutchings et al., 2001). The nitrate leaching is, for all methodologies, calculated as a proportion of N input in fertilisers, manure and BNF. The standard proportion of N input that is leached was set to 30% (IPCC, 1997). Methane from faeces deposited on grazed grassland is estimated according to IPCC (1997). Nitrous oxide emissions are estimated as a proportion of all N inputs to the field, including crop residues and BNF, but excluding atmospheric N deposition. The emission factor generally used is 1.25% (IPCC, 1997, 2000). However, an emission factor of 2% is used for N deposited by grazing animals. The emission factor is 1% for N lost by NH3 volatilisation, and 2.5% for N lost by nitrate leaching. 2.1.5. Pre-chain emissions The pre-chain emissions are the emissions associated with imports of production goods to the farm. In the model this includes consumption of energy, fertilisers, pesticides and feedstuffs. However, energy costs represented by buildings and farm machinery are not considered by the model. Two types of energy sources are used by the model farms, diesel and electricity. The emissions associated with the

production, transport and use of these two energy sources were taken as values being representative for Central Europe. The diesel use associated with different field operations are specified in the model, as is the use of electricity associated with housing, feeding and supply of water for drinking and irrigation (Bockisch, 2000; Dalgaard et al., 2002). The average GHG emissions associated with the supply of mineral N fertiliser representative for the conditions in Germany (transportation distances, energy mix, etc.) are used in the model (Patyk and Reinhardt, 1997). The emissions also include N2O from fertiliser production. The emissions associated with pesticide consumption are taken as the emissions associated with the formulation and packaging of an average pesticide (Green, 1987; Kaltschmitt and Reinhardt, 1997). For the import of seeds and feed, the emissions are calculated as the sum of emissions from energy use (Kaltschmitt and Reinhardt, 1997; Bockisch, 2000), and from N2O emissions associated with fertiliser use and BNF using standard yields and N inputs. 2.2. European model dairy farms Five European dairy farming regions were identified to reflect differences in livestock density and differences in grazing/feeding systems; yet, each region covers large variations in dairy farm management (European Commision, 2000). In the Atlantic zone, which has been defined as the region with the highest intensity of dairy production, grazing systems dominate. The Continental region also encompasses areas with high stocking rates, but feeding systems often include fair amounts of maize silage. The Boreal, the Pre-Alpine and the Mediterranean regions have a lower stocking rate. Boreal systems typically include grazing, whereas grazing is less common in the Pre-Alpine and Mediterranean systems. Within each region, sets of conventional and organic farms were defined (Table 2). No attempt was made to define truly representative farms of the region. Instead, farms were defined based on selected experimental or prototype farms with generally good management. The climate, soil type and atmospheric N deposition of the model farms were also taken as those representing these experimental farms (Table 3). The model dairy farms were all defined to have the same utilised agricultural area (50 ha). Cows on conventional and organic model farms were defined to achieve the same milk yield, so the basic difference between conventional and organic farms was expressed in the livestock density. The organic farms were defined to be 100% self-sufficient with respect to feed, meaning that the livestock density depends solely on the feed that can be produced off the farmland. The conventional farms, on the other hand, import concentrates as supplementary feed, and their livestock density was defined to be 75% higher than the density on the

Table 2 Basic characteristics for the 15 model dairy farms, each 50 ha, in five European regions Atlantic

a b c d e f g h

Pre-Alpine

Boreal

Mediterranean

1 Conventional Mixed Slurry 7000

2 Organic Mixed Slurry 7000

3 Conventional Grass Slurry 7000

4 Conventional Maize Slurry 7000

5 Conventional Mixed Slurry 6000

6 Organic Mixed Slurry 6000

7 Organic Mixed Deep litter 6000

8 Conventional Mixed FYM 5500

9 Organic Mixed Slurry 5500

10 Organic Mixed FYM 5500

11 Conventional Mixed Slurry 7000

12 Organic Mixed Slurry 7000

13 Conventional Mixed Slurry 5500

14 Organic Mixed Slurry 5500

15 Organic Mixed FYM 5500

73/76

42/44

73/76

73/76

60/63

34/36

34/36

58/61

33/35

33/35

31/32

18/18

68/71

39/41

39/41

2.7 217

1.5 140

2.7 214

2.7 224

2.2 174

1.2 114

1.2 118

2.1 148

1.2 112

1.2 105

1.1 86

0.7 48

2.5 186

1.4 137

1.4 128

D 180

D 180

DN 180

DN 180

D 163

D 163

D 163

– 158

D 158

D 158

– 123

– 123

– 168

– 168

– 168

30 42 28 0 0 382

30 56 14 0 0 365

40 60 0 0 0 365

40 10 50 0 0 416

22 22 14 42 0 369

24 41 0 28 7 274

24 41 0 28 7 274

20 48 8 16 8 333

30 46 0 16 8 254

30 46 0 16 8 254

6 19 0 75 0 215

6 19 0 75 0 136

22 46 16 16 0 381

23 37 13 27 0 325

23 37 13 27 0 325

FYM is the separate system with both solid and liquid manure. Livestock units, one dairy cow is 1.2 LU and 1 heifer is 0.6 LU. Including manure deposited by cattle on grazed areas. DN is day and night time grazing, D is daytime grazing, (–) indicates no grazing. Grazing or fresh feed inside. Crop area in percent of farm area. Grass, clover and grain crops for silage; alfalfa for hay. DM losses in grazing, silage making and harvesting have been taken into account.

J.E. Olesen et al. / Agriculture, Ecosystems and Environment 112 (2006) 207–220

Farm number Farm type Crop rotation Manure type a Milk (kg cow1 year1) Herd size (cows/ young stock) LU (ha1) b Livestock manure (kg N ha1 year1)c Grazingd Grazing/summer feeding (days) Fresh grass (%)e,f Forage crops (%)g,f Maize silage (%) f Grain crops (%) f Potatoes (%) f Total net yield (t DM year1)h

Continental

213

214

J.E. Olesen et al. / Agriculture, Ecosystems and Environment 112 (2006) 207–220

Table 3 Location of the site used to represent climate (range of mean monthly temperatures) and atmospheric N deposition for the model farms in the five European regions Region

Location

Mean temperature range (8C)

N deposition (kg N ha1 year1)

Atlantic Continental Pre-Alpine Boreal Mediterranean

The Netherlands Central Germany Austria Southern Finland Po Valley, Italy

2.5–17.1 0.2–17.5 1.6–17.9 7.3–15.8 1.3–25.0

29 20 24 7 19

basic settings, an annual stock replacement with younger cows of 40% was assumed. Data for mean monthly normal climate (temperature) were taken from the climatological station nearest to the prototype farm. The range of mean monthly temperatures is shown in Table 3 along with the estimated N deposition at these sites (Fagerli et al., 2003).

corresponding organic farm. Regional differences between farms were expressed in the milk yield, the crop rotation systems, and the cow housing system and manure management method most common to the region. The crop yields were estimated based on experience within the regions, and the crop N demands were adjusted to match these yields (Schelde et al., 2004). Systems for feed evaluation and recommendations for ration formulation differ within Europe, and no common European approach exists (Kaustell et al., 1997). It was not feasible to use separate feed evaluation systems for each region/country, so the system used in Denmark was applied for all regions (LR, 1999; Møller et al., 2000). Annual feed plans were formulated on the condition that all available crops on farm (except for potatoes) were used for animal feed. For organic farms, this typically resulted in a feed plan dominated by forage crops. For the conventional farms, a customised mixture of concentrates was imported to reach a feed plan with a prescribed protein balance. In the model farms it was pre-conditioned (for ease of calculations) that all bull calves are exported from the dairy model farm as newborn. Thus, only dairy cows and heifers are accounted for in the feed requirement calculations. In the

3. Results 3.1. Variation in emission estimates The model was applied to the European dairy model farms shown in Table 2. Nitrous oxide contributed on average about 49% of the total emissions in terms of global warming potential, and CH4 contributed about 42% (Table 4). The animals contributed about 36% and the fields about 39% to the total emissions (data not shown). The uncertainty in simulated emissions was estimated by applying both the IPCC (1997) tier 1 methodology, the IPCC (2000) tier 2 methodology and the default FarmGHG methodology (Table 4). The standard deviation for the

Table 4 Simulated emissions of CO2, CH4 and N2O (Mg CO2-eq ha1 year1) using three different methods for estimating greenhouse gas emissions, the default FarmGHG methodology, the IPCC (1997) tier 1 methodology and the IPCC (2000) tier 2 methodology Farm no.

Type

CO2

CH4

Default

IPCC 1

IPCC 2

Default

IPCC 1

N2O IPCC 2

Default

1 2 3 4

Conventional Organic Conventional Conventional

1.5 0.4 1.9 1.3

5.9 3.6 5.6 5.6

8.9 6.1 7.8 7.8

5.6 4.7 5.4 5.2

4.1 1.8 4.9 4.2

5.0 3.0 6.4 5.1

6.5 3.4 8.6 6.3

5 6 7

Conventional Organic Organic

0.9 0.3 0.3

4.5 2.6 2.2

5.3 3.3 2.1

3.8 2.6 2.4

3.5 1.6 2.4

4.3 2.5 3.3

4.8 2.9 3.2

8 9 10

Conventional Organic Organic

1.1 0.3 0.3

3.7 2.6 2.1

3.4 3.2 2.0

3.5 2.6 2.3

3.2 1.5 1.8

4.2 2.5 2.7

4.8 2.9 2.9

11 12

Conventional Organic

0.7 0.3

2.4 1.4

3.2 1.8

2.0 1.1

2.1 0.8

2.1 1.0

2.8 1.6

13 14 15

Conventional Organic Organic

1.1 0.4 0.4

5.6 3.4 2.5

7.6 5.1 2.7

5.5 4.5 3.4

2.5 1.6 2.1

3.6 2.6 3.0

6.1 3.8 4.0

J.E. Olesen et al. / Agriculture, Ecosystems and Environment 112 (2006) 207–220

215

Fig. 2. Comparison of simulated and measured N2O emissions from fields of pairs of organic and conventional farms using measurements from crop rotations on prototype farms (Petersen et al., 2004). The coefficient of determination (R2) of the relationship is 0.48.

different methods was about 19% of the respective mean emission values for both CH4 and N2O. The N2O emissions were in general higher with the default FarmGHG methodology compared with the IPCC tier 2 methodology, whereas the CH4 emissions for most farms were lower with the default FarmGHG methodology compared with the IPCC tier 2 methodology. For some of the prototype farms, on which these model farms were based or which had similar characteristics, N2O emissions were monitored in organic and conventional crop rotations (Petersen et al., 2004), thus allowing comparison of observed and simulated field N2O emissions (Fig. 2). The measurements were taken on eight farms in Austria, Finland, Italy and the UK. There was, in particular for the conventional crop rotations, an excellent agreement between simulated and observed emissions. 3.2. Relationships with N cycling The farm N surplus was calculated as the difference between imported and exported N. The imported N included imported fertilisers, feed, seeds, N deposition and BNF, and the exported N included N in meat, milk and plant products sold from the farm. The N surplus was found to increase with increasing livestock density on both conventional and organic dairy farms (Fig. 3a). The simulated N efficiencies, calculated as the ratio of exported over imported N, generally varied between 16 and 26%, and there was a tendency for a decline with increasing livestock density (Fig. 3b). However, the N efficiencies for the Mediterranean farm were considerably lower (12–15%), which can be attributed to the large estimated BNF in the alfalfa crop of these farms.

Fig. 3. Farm N surplus calculated as imported minus exported N (a) and N use efficiency calculated as exported over imported N (b) depending on livestock density. The coefficient of determination of the relationships (R2) is indicated.

The GHG emissions increased with increasing N surplus, and on an area basis there was no difference in the linear relationship between conventional and organic farms (Fig. 4a). The slope of the regression line indicates an increase in GHG emissions of 0.76 Mg CO2-eq kg1 N in N surplus. There was a tendency for higher emissions per unit farm area from conventional compared with organic farms for similar farm N efficiencies (Fig. 4b). When calculating the emissions on the basis of milk production, organic farms tended to have higher emissions than conventional farms at similar farm N surplus and N efficiency (Fig. 4c and d). The farms also exported meat and some plant products (e.g. cereals and potatoes), and Fig. 4e and f shows the GHG emissions per unit of energy in the exported produce. Here, there was a linear decline in emissions per energy unit with increasing farm N efficiency (Fig. 4f), and the slope corresponded to 16.5 kg CO2-eq MJ1 per % increase in efficiency. 3.3. N2O emissions The major part of N2O emissions originated from the fields, but the distribution among sources of N2O varied

216

J.E. Olesen et al. / Agriculture, Ecosystems and Environment 112 (2006) 207–220

Fig. 4. Farm greenhouse gas emissions depending on farm N surplus (a, c and e) and on farm N efficiency (b, d and f). The emissions are shown as emissions per farmed area (a and b), emissions per kg milk produced (c and d), and emissions per MJ of metabolic energy in the exported milk, meat and plant products. The coefficient of determination of the relationships (R2) is indicated.

somewhat between the different farm types (Fig. 5). Emissions from fertiliser and manure were considerably higher on conventional farms compared with organic farms, whereas emissions derived from BNF were highest on the organic farm (Fig. 5a). There were minor variations among different conventional farm types in the Atlantic region (Fig. 5b). However, the farm with the grass rotation had a higher fertiliser N import, resulting in higher N leaching compared with the other farms. The indirect N2O emissions from N leaching constituted a rather large proportion of total N2O emissions. This is partly due to the high N2O emission factor of 0.025 used for N leaching, and partly because a constant fraction (30%) of N-input was assumed to be lost by leaching. The estimated N leaching constituted 53 and 66% of field N surplus for

conventional and organic farms, respectively. When the estimated NH3 volatilisation was subtracted from the field N surplus, N leaching constituted 66 and 84% of the corrected N surplus for conventional and organic farms, respectively. 3.4. CH4 emissions Methane emissions primarily originated from enteric fermentation and from manure storage. Enteric fermentation was almost constant over the year with some variation between winter and summer due to different feeding regimes (Fig. 6). Although the livestock density and thus the milk production on the conventional farms was defined to be 75% above corresponding conventional farms, the increase in CH4 emissions from enteric fermentation was only 9–81%

J.E. Olesen et al. / Agriculture, Ecosystems and Environment 112 (2006) 207–220

217

between livestock density and N surplus shown in Fig. 3a is in line with the relationship observed for organic farms and some conventional farms in Denmark (Dalgaard et al., 1998; Halberg, 1999). This probably reflects the assumption of good farm management practices on the model farms, which has the effect of reducing unnecessarily high N fertiliser inputs, and thus the N surplus, on the conventional farms. The simulated N efficiencies varied between 16 and 26%, when Mediterranean farms were excluded. This is lower than the N efficiencies of 24–29% found by Swensson (2003) for dairy farms in Sweden. However, Swensson (2003) did not include atmospheric N deposition in the estimates. Dalgaard et al. (1998) found average N efficiencies of 18–20% for conventional, and 25–28% for organic dairy farms in Denmark, whereas Halberg (1999) did not find a significant difference in N efficiency between organic and conventional dairy farms in Denmark. The simulated N efficiencies were mostly within this range, and thus in agreement with observed N efficiencies of dairy farms in Northern Europe. The low N efficiencies for the Mediterranean farms can be attributed to the large estimated BNF in the alfalfa crop of this farm type. There is some uncertainty associated with these estimates, since manure was also applied to some of the alfalfa crops. Fig. 5. Sources of nitrous oxide emissions from the fields on the farms in the Atlantic region, including comparisons of a conventional and organic farm (farms 1 and 2) (a) and comparisons of different crop rotations (farms 1, 3 and 4) (b).

(average 36%) for conventional compared with organic farms, indicating a large influence of feeding practice on emissions. The farms shown in Fig. 6 are all slurry-based and have higher emissions from the manure handling during summer than during winter, in particular for the Mediterranean farms.

4. Discussion 4.1. Farming intensity Greenhouse gas emissions per land area increased with increasing N surplus (Fig. 4a). This is a consequence of the link between N surplus and livestock density, and thus with production intensity. Also, the high contribution of N2O to total greenhouse gas emissions means that a more intense and leaky N cycle will significantly increase total greenhouse gas emissions. In addition, CH4 emissions increase with livestock density due to a fairly constant emission per animal, depending on feeding regime. The linear increase in farm N surplus with increasing livestock density was similar for organic and conventional dairy farms (Fig. 3a). This is in contrast to Dalgaard et al. (1998), who found the N surplus from organic farms to be substantially smaller than for conventional farms, when a correction for livestock density was made. The relationship

4.2. Farm N efficiency The simulated greenhouse gas emissions varied between 1.2 and 1.7 kg CO2-eq kg1 milk for the conventional farms (Fig. 4c and d). This is somewhat higher than the estimate of 1.09 kg CO2-eq kg1 milk estimated for dairy systems in the USA (Phetteplace et al., 2001). This could indicate more efficient dairy production systems in the USA. However, the difference could also be due to differences in the methodology applied, including emission factors for CH4 and N2O and the accounting of pre-chain emissions, or to uncertainties in the budgets of imports and exports from the dairy farms (Oenema et al., 2003). There is thus a need for intercomparison of farm scale models of greenhouse gas emissions to reveal differences and uncertainties. There was a significant (P < 0.01) decline in GHG emissions per kg milk with increasing farm N efficiency (Fig. 4d). The relationship was even stronger (P < 0.0001) when the GHG emission were expressed per MJ of energy in the exported products (Fig. 4f). The better relationship in the latter case is probably an effect of a generally higher N use efficiency in plant production compared with animal production, because of more N loss pathways associated with manure management and a smaller export of N in animal products (Bleken and Bakken, 1997; Schro¨der et al., 2003). The close relationship between GHG emissions per product unit and farm N efficiency indicates that the farm N efficiency could be used as a proxy for comparing the efficiencies of farms with respect to supplying products with

218

J.E. Olesen et al. / Agriculture, Ecosystems and Environment 112 (2006) 207–220

Fig. 6. Monthly methane emissions from farms with slurry based manure management systems for conventional (left column) and organic farms (right column) in the Atlantic, Continental, Boreal and Mediterranean regions.

a low GHG emission. The relationship was strongly significant for both organic and conventional farms, but with slightly higher GHG emissions from organic farms at similar farm N efficiency. This may partly be due to the higher estimated CH4 emissions from enteric fermentation due to a higher proportion of forage crops in the diet. This would indicate that organic farms might not have an advantage from a GHG emissions perspective. However,

these results are based on model farms with best management practices, and there is a need to test whether this trend also holds true for a comparison of real organic and conventional farms, where the management is often suboptimal. Additionally, the N2O emissions derived from N leaching from the organic farms may have been overestimated, since the ratio of N leaching to total farm N surplus was higher for organic farms than for conventional

J.E. Olesen et al. / Agriculture, Ecosystems and Environment 112 (2006) 207–220

farms. The N leaching in FarmGHG was estimated as a fixed proportion of field N input. However, in practice a non-linear relationship exists between applied N and N leaching (Simmelsgaard and Djurhuus, 1998), giving relatively higher N leaching at higher N inputs.

219

CH4 emissions is needed. For the manure management, particular emphasis should be put on improving our understanding of CH4 emissions from the entire management chain, and for fields a better estimation of N2O emissions along with other N losses is needed.

4.3. Farm type 5. Conclusions The simulated annual direct N2O emissions from the fields in the Atlantic region varied from 5.2 to 6.9 kg N2ON ha1. Brown et al. (2001) estimated N2O emissions for two dairy farms in UK using a different methodology with estimates varying from 6.9 to 8.2 N2O-N ha1, which is of the same magnitude as found in this study. The agreement between simulated results and measurements of N2O emissions at the crop rotation level (Fig. 2) and measurements of CH4 emissions at the farm scale (data not shown) shows that the model simulates the level of the observed emissions and further replicates a large proportion of the variation between farms and crop rotations. There was some scatter between observed and simulated emissions from organic farms, which may be related to the type of manure being applied on some of these farms (Petersen et al., 2004). Despite the general agreement between model and measurement results, there is still considerable need to improve the methodology for estimating CH4 and N2O emissions from dairy farms. 4.4. Model improvement In many cases, the model makes use of empirical relationships derived in a single country in Europe. In particular the feed evaluation system, fertiliser replacement values and NH3 emission factors were taken as the standard values used in Denmark, because there are no unified European approaches that have the sufficient detail. For the feed evaluation system this does probably not affect results, since the model assumes that a Holstein-Frisian breed is used on all farms. However, the NH3 emission factors from the housing and manure storages may have been underestimated for Southern European conditions, where temperatures are higher. For the same reason, the fertiliser replacement values may be overestimated in Southern Europe. However, this would depend on the actual timing of manure applications. There is a need to develop a model that includes also the effect of climatic variation on these factors. Even though the model applies the current state-of-art in the understanding of emissions from manure management systems, these emissions are still very uncertain, as illustrated by large differences between similar farms with different manure management practices (Table 4). More data, and a better understanding of the importance of various processes in the farm flow chains, are still needed. This may be illustrated by the emissions from deep litter mats, where there are no credible data available. For the animals, a verification of the effects of varying diet composition on

The FarmGHG model was used to simulate farm N flows, N surpluses and greenhouse gas emissions for a set of European model dairy farms that are broadly in agreement with observations from conventional and organic dairy farms in Northern Europe. The simulated total GHG emissions could be closely related to either the farm N surplus or the farm N efficiency. The farm N surplus appears to be a good proxy for GHG emissions per unit of land area. The farm N surplus can relatively easily be determined on practical farms from farm records of imports and exports and the composition of the crop rotation. The relationship shown here, however, needs to be verified for real dairy farming systems across Europe. If this relationship is validated, it would provide a simple alternative approach to perform inventories of GHG emissions from farming systems. The farm N efficiency appears to be a good proxy for emissions per unit of production. It implies that a reduction in GHG emissions from agricultural production can be achieved by increasing the N efficiency of the agricultural sector. However, the N efficiency is strongly affected by the ratio of crop to animal products in the farm output, and further studies are needed to clarify the relationship for different farm types.

Acknowledgement The work was supported by the European Union under contract EVK-CT2-2000-00096.

References Amon, B., 1999. NH3-, N2O- und CH4-emissionen aus der Festmistanbindehaltung fu¨r Milchvieh–Stall–Lagerung–Ausbringung. Dissertation. Forschungsbericht Agrartechnik 331. Universita¨t fu¨r Bodenkultur, Institut fu¨r Land-, Umwelt und Energietechnik, Wien, Austria. Amon, B., Kryvotuchko, V., Amon, T., Be´line, F., Petersen, S.O., 2004. Quantitative Effects of Storage Conditions on GHG Emissions from Cattle Slurry, and N2O and CH4 Turnover Inside Natural Surface Crusts. Final Report MIDIAIR Project. Institute for Energy and Environment, Leipzig, Germany. Bleken, M.A., Bakken, L.R., 1997. The nitrogen of food production: Norwegian study. Ambio 26, 134–142. Bockisch, F.-J., 2000. Bewertung von Verfahren der o¨kologischen und konventionellen landwirtschaftlichen Produktion im Hinblick auf den Energieeinsatz und bestimmte Schadgasemissionen. Sonderheft 211. Bundesanstalt fu¨r Landwirtschaft, Wissenschaftliche Mitteilungen der

220

J.E. Olesen et al. / Agriculture, Ecosystems and Environment 112 (2006) 207–220

Bundesanstalt fu¨r Landwirtschaft, Landbauforschung Vo¨lkenrode, Germany, 209 pp. Brown, L., Jarvis, S.C., Headon, D., 2001. A farm-scale basis for predicting nitrous oxide emissions from dairy farms. Nutr. Cycl. Agroecosyst. 60, 149–158. Dalgaard, T., Halberg, N., Kristensen, I.S., 1998. Can organic farming help to reduce N-losses. Nutr. Cycl. Agroecosyst. 52, 277–287. Dalgaard, T., Dalgaard, R., Nielsen, A.H., 2002. Energiforbrug pa˚ økologiske og konventionelle landbrug. Grøn Viden, Markbrug no. 260. Danish Institute of Agricultural Sciences, Tjele, Denmark, 8 pp. European Commision, 2000. The Environmental Impact of Dairy Production in the EU: Practical Options for the Improvement of the Environmental Impact. Report Prepared by CEAS Consultants Ltd., Centre for European Agricultural Studies, 190 pp., available at http://europa.eu.int/comm/ environment/agriculture/pdf/dairy.pdf (accessed on 20 July 2005). Fagerli, H., Simpson, D., Tsyro, S., Solberg, S., Aas, W., 2003. Transboundary Acidification, Eutrofication and Ground Level Ozone in Europe. Part II. Unified EMEP Model Performance. EMEP Status Report 2003. Norwegian Meteorological Institute, Oslo, Norway, 170 pp. Green, M.B., 1987. Energy in pesticide manufacture, distribution and use. In: Helsel, Z.R. (Ed.), Energy in Plant Nutrition and Pest Control. Energy in World Agriculture, 2. Elsevier, Amsterdam, pp. 165–177. Halberg, N., 1999. Indicators of resource use and environmental impact for use in a decision aid for Danish livestock farmers. Agric. Ecosyst. Environ. 76, 17–30. Houghton, J.T., Ding, Y., Griggs, D.J., Noguer, M., van der Linden, P.J., Dai, X., Maskell, K., Johnson, C.A., 2001. Climate Change 2001: The Scientific Basis. Contribution of Working Group I to the Third Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, UK, 881 pp. Husted, S., 1994. Seasonal variation in methane emission from stored slurry and solid manures. J. Environ. Qual. 23, 585–592. Hutchings, N.J., Sommer, S.G., Andersen, J.M., Asman, W.A.H., 2001. A detailed ammonia emission inventory for Denmark. Atmos. Environ. 35, 1959–1968. Hu¨ther, L., 1999. Entwicklung analytischer Methoden und Untersuchung von Einflußfaktoren auf Ammoniak-, Methan- und Distickstoffmonoxidemissionen aus Flu¨ssig- und Festmist. Landbauforschung Vo¨lkenrode Sonderheft 200, FAL, Braunschweig, Germany, 225 pp. Høgh-Jensen, H., Loges, R., Jørgensen, F.V., Vinther, F.P., Jensen, E.S., 2004. Empirical model for quantification of symbiotic nitrogen fixation in leguminous crops. Agric. Syst. 82, 181–194. IPCC, 1997. Greenhouse Gas Inventories. Revised 1996 IPCC Guidelines for National Greenhouse Gas Inventories. OECD, Paris. IPCC, 2000. IPCC Good Practice Guidance and Uncertainty Management in National Greenhouse Gas Inventories. OECD, Paris. Kaltschmitt, M., Reinhardt, G.A., 1997. Nachwachsende Energietra¨gerGrundlagen, Verfahren, o¨kologische Bilanzierung. Vieweg-Verlag, Braunschweig/Wiesbaden, Germany. Kaustell, K., Tuori, M., Huhtanen, P., 1997. Comparison of energy evaluation systems for dairy cow feeds. Livest. Prod. Sci. 51, 255–266. Kirchgessner, M., Windisch, W., Mu¨ller, H.L., 1995. Nutritional factors for the quantification of methane production. In: Engelhardt, W.v., Leonhard-Marek, S., Breves, G., Giesecke, D. (Eds.), Ruminant Physiology: Digestion, Metabolism, Growth and Reproduction. Proceedings VIII International Symposium on Ruminant Physiology. pp. 333–348. LR, 1999. Ha˚ndbog i kvæghold. The Danish Agricultural Advisory Centre, Aarhus, Denmark. LR, 2002. Ha˚ndbog i plantedyrkning. The Danish Agricultural Advisory Centre, Aarhus, Denmark. Monteny, G.J., Groenestein, C.M., Hilhorst, M.A., 2001. Interactions and coupling between emissions of methane and nitrous oxide from animal husbandry. Nutr. Cycl. Agroecosyst. 60, 123–132. Mosier, A.R., Duxbury, J.M., Freney, J.R., Heinemeyer, O., Minami, K., Johnson, D.E., 1998a. Mitigating agricultural emissions of methane. Clim. Change 40, 39–80.

Mosier, A., Kroeze, C., Nevison, C., Oenema, O., Seitzinger, S., van Cleemput, O., 1998b. Closing the global N2O budget: nitrous oxide emissions through the agricultural nitrogen cycle. Nutr. Cycl. Agroecosyst. 52, 225–248. Møller, J., Thøgersen, R., Kjeldsen, A.M., Weisbjerg, M.R., Søegaard, K., Hvelplund, T., Børsting, C.F., 2000. Feedstuff Table. Composition and Feeding Value of Feedstuffs for Cattle. Report No. 91. The Danish Agricultural Advisory Centre, Aarhus, Denmark, 52 pp. Oenema, O., Gebauer, G., Rodriguez, M., Sapek, A., Jarvis, S.C., Corre´, W.J., Yamulki, S., 1998. Controlling nitrous oxide emissions from grassland livestock production systems. Nutr. Cycl. Agroecosyst. 52, 141–149. Oenema, O., Kros, H., de Vries, W., 2003. Approaches and uncertainties in nutrient budgets: implications for nutrient management and environmental policies. Eur. J. Agron. 20, 3–16. Olesen, J.E., Weiske, A., Asman, W.A.H., Weisbjerg, M.R., Djurhuus, J., Schelde, K., 2004. FarmGHG. A Model for Estimating Greenhouse Gas Emissions from Livestock Farms. Documentation. DJF Internal Report No. 202. Danish Institute of Agricultural Sciences, Tjele, Denmark, 54 pp. Patyk, A., Reinhardt, G.A., 1997. Du¨ngemittel-Energie-und Stoffstrombilanzen. Vieweg-Verlag, Braunschweig/Wiesbaden, Germany. Petersen, S.O., Regina, K., Po¨llinger, A., Rigler, E., Valli, L., Yamulki, S., Esala, M., Fabbri, C., Syva¨salo, E., Vinther, F.P., 2004. Nitrous oxide emissions from organic and conventional crop rotations in five European countries. Agric. Ecosyst. Environ. (this issue). Phetteplace, H.W., Johnson, D.E., Seidl, A.F., 2001. Greenhouse gas emissions from simulated beef and dairy livestock systems in the United States. Nutr. Cycl. Agroecosyst. 60, 99–102. Poulsen, H.D., Børsting, C.F., Rom, H.B., Sommer, S.G., 2001. Kvælstof, fosfor og kalium i husdyrgødning–normtal 2000. DJF Report No. 36. Danish Institute of Agricultural Sciences, Tjele, Denmark, 152 pp. Schelde, K., Olesen, J.E., Weisbjerg, M.R., 2004. MIDAIR. Definition of Conventional and Organic Dairy Farms in Five European Regions. DJF Internal Report No. 206. Danish Institute of Agricultural Sciences, Tjele, Denmark, 69 pp. Schro¨der, J.J., Aarts, H.F.M., ten Berge, H.F.M., van Keulen, H., Neeteson, J.J., 2003. An evaluation of whole-farm nitrogen balances and related indices for efficient nitrogen use. Eur. J. Agron. 20, 33–44. Simmelsgaard, S.E., Djurhuus, J., 1998. An empirical model for estimating nitrate leaching as affected by crop type and the long-term N fertilizer rate. Soil Use Manage. 14, 37–43. Sommer, S.G., 2001. Effect of composting on nutrient loss and nitrogen availability of cattle deep litter. Eur. J. Agron. 14, 123–133. Sommer, S.G., Petersen, S.O., Søgaard, H.T., 2000. Greenhouse gas emission from stored livestock slurry. J. Environ. Qual. 29, 744–751. Swensson, C., 2003. Analyses of mineral element balances between 1997 and 1999 from dairy farms in the south of Sweden. Eur. J. Agron. 20, 63– 69. Søgaard, H.T., Sommer, S.G., Hutchings, N.J., Huijsmans, J.F.M., Bussink, D.W., Nicholson, F., 2002. Ammonia volatilization from field applied animal slurry—the ALFAM model. Atmos. Environ. 36, 3309– 3319. Velthof, G.L., van Beusichem, M.L., Oenema, O., 1998. Mitigation of nitrous oxide emission from dairy farming systems. Environ. Poll. 102, 173–178. Watson, C.A., Atkinson, D., 1999. Using nitrogen budgets to indicate nitrogen use efficiency and losses from whole farm systems: a comparison of three methodological approaches. Nutr. Cycl. Agroecosyst. 53, 259–267. Weisbjerg, M.R., Hvelplund, T., 1993. Bestemmelse af nettoenergiindhold (FEK) i ra˚varer og kraftfoderblandinger. Forskningsrapport nr. 3 National Institute of Animal Science, Tjele, Denmark. Weiske, A., Vabitsch, A., Olesen, J.E., Schelde, K., Michel, J., Friedrich, R., Kaltschmitt, M., 2005. Mitigation of greenhouse gas emissions in European conventional and organic dairy farming. Agric. Ecosyst. Environ. (this issue).