Interactive comment on “ Estimation of nitrogen budgets for contrasting catchments at the landscape scale ” by E

The authors thank the Anonymous Reviewer for their comments. We have responded to the specific comments below. (Note: "Author’s reply:..." inserted after the original text from the Review ("Referee #1:...") Referee #1: I do not feel comfortable with the concept “landscape NANI” proposed by the authors in this paper. The strength of the NANI approach (recently overviewed by Swaney et al. 2012) lies in the fact that only “new nitrogen” is accounted. With this approach all double counting related to N recirculation is avoided. On the other hand NANI approach has not the power of resolution achieved by authors in this work.


Introduction
Human activities dominate the global nitrogen (N) budget by adding reactive forms of nitrogen (N r ) to the environment (Galloway et al., 2004). The main forms of anthro- 25 pogenic N r are reduced (e.g. NH 3 , NH + 4 ), oxidised (e.g. NO 2 , N 2 O, NO − 3 ) and organic 8990 effective regulations aimed at reducing environmental impacts. However, accurate estimation of N fluxes at high spatial resolution poses a significant challenge (de Vries et al., 2011), e.g. the estimation of spatially variable N dry deposition represents one of the key uncertainties in quantifying nitrogen inputs to terrestrial ecosystems  with a detailed spatial landscape inventory of field specific agricultural activities. The study shows how landscape N budget analysis provides an insight into the main N flux terms, key uncertainties associated with these terms and the overall implications for the environmental status of the landscape.

Study landscape
As part of the NitroEurope Integrated Project (Sutton et al., 2007), a landscape study area of 6 km × 6 km was established in southeast Scotland, an area with a temperate oceanic climate, for detailed inventory of agricultural activities, N r concentration and flux measurements (see Vogt et al., 2012a, b, for further details). The study landscape 10 was located to include the two contrasting catchments. The moorland peat-dominated catchment covered 621 ha, while the grassland dominated catchment covered 895 ha. Together these two catchments represent 42 % of the study landscape ( Fig. 1). A detailed local survey of all farms and fields in the study landscape was conducted throughout 2008. This provided land cover and farm activity data, which were collated 15 into a relational database and spatially represented in a geographical information system (ArcGIS, ESRI). Land cover and soil types within the landscape together with the boundaries of the two studied catchments are shown in Fig. 1. Moorland and rough grass, including peat cutting and areas of both deciduous and coniferous afforestation dominate the northwestern part of the landscape and the Black Burn catchment, 20 whereas the southeast and the Lead Burn catchment is dominated by agricultural land (henceforth referred to as the Moorland and the Grassland catchments, respectively, Table 1). Agricultural activities in the landscape range from extensive beef cattle and sheep farming to intensive poultry farming, with 24 poultry houses in the study area containing nearly 1. 5

Catchment N budgets
Two recent studies have compared different budgetary approaches to quantify N balances in agricultural systems. Oenema et al. (2003) presented farm gate, soil surface and soil system budget methodologies, and de Vries et al. (2011) presented regional farm, land and soil N budgets, both studies delineated inputs and outputs for 5 each of these approaches. In our study, the Moorland and Grassland catchment annual N budgets were assessed for 2008 using a soil budgeting approach which most closely matches the soil N budget approach of de Vries et al. (2011) (except the N pool changes), i.e. all N that enters and leaves the soil was accounted for. This type of approach was chosen as the inputs and outputs are directly associated with the catch-10 ment soils and linked to the downstream flux. The balance of the N input and output terms indicate the change in N storage within the catchment over time. There were significant N fluxes occurring in connection with the poultry housing, i.e. housing emissions and farming operations such as feed import, manure export or livestock export within one of the catchments, but for the purpose of a soil budget approach, housing 15 emissions and farming operations not affecting the catchment land surface were considered decoupled from the soil. Thus they were excluded from this approach, except via the N deposition fluxes resulting from housing emissions. The soil N budget was derived as follows: where ∆N/∆t is the change in N balance (∆N) over time (∆t); N NH 3 dry dep is the atmospheric dry deposition of ammonia (NH 3 ); N NH x wet dep is the atmospheric wet deposition of reduced nitrogen (NH x ); N NO y dep is the atmospheric dry and wet deposition of 25 oxidised nitrogen (NO y ); N syn fert is the N content in applied synthetic fertiliser; N org fert is the N content in applied organic fertiliser; N excreta is the amount of N excreted by grazing livestock; N bio fix is the biological N 2 fixation; N NH 3 , N of NH 3 , nitrous oxide (N 2 O), nitric oxide (NO) and N 2 to the atmosphere; N harvest is the N offtake through harvested vegetation for silage and hay production; N grass is the N offtake through harvested grass by grazing livestock; N stream is the downstream export flux of total dissolved nitrogen (TDN). The uncertainties of individual budget terms are given by estimated positive and 5 negative errors (Sect. 3.7). The overall uncertainty of the N balance (E ∆N/∆t ) was calculated as the square root of the sum of the error (E ) squares, hereby accounting for the depending variables N grass and N excreta : In the following sections the method of quantifying individual budget terms and their uncertainties is described.

Atmospheric deposition
The spatial and temporal variability of atmospheric NH 3 across the landscape, in which the two catchments are contained, was described in detail by Vogt et al. (2012b). Monthly mean NH 3 concentrations at 31 sites were measured through 2008 with ALPHA passive diffusion samplers (Tang et al., 2001). Sites were distributed across 20 the study landscape with an emphasis on capturing high and low emission areas as well as the variability around sources. Ammonia emissions were calculated for each individual field, manure store and livestock house, based on the field and farm activities recorded on a monthly basis combined with emission rates for each activity (manure housing, storage and spreading, grazing and fertiliser application, Vogt et al., 2012b). 25 The emission estimates were used in the Local Area Dispersion and Deposition model 8994 Introduction  (Hill, 1998;Loubet et al., 2009) at a resolution of 25 m x 25 m to model spatial concentrations and dry deposition of NH 3 within the study landscape. Measured annual mean concentrations of the 31 sampling sites were used for verification of the LADD model. As NH 3 has a high dry deposition rate (Cellier et al., 2011) and is thus expected to be driven by local sources, NH 3 dry deposition inputs to the studied catch-5 ments (N NH 3 dry dep ) were calculated from fluxes modelled by LADD within the study landscape only (accounting for atmospheric NH 3 import to the landscape using national modelling). This N budget term is considered to carry a relatively low uncertainty of ±20 % in this instance due to the detailed local study, involving an intensive measurement programme and local atmospheric dispersion modelling.
10 Catchment atmospheric inputs due to NH x wet deposition (N NH x wet dep ) and dry and wet deposition of NO y (N NO y dep ) which are expected to be largely driven by non-local sources (e.g. Hertel et al., 2011;Sutton et al., 1998) were simulated by the UK national model FRAME (Fine Resolution Atmospheric Multi-pollutant Exchange) (Dore et al., 2012(Dore et al., , 2007Hallsworth et al., 2010) at a resolution of 1 km × 1 km. The contribution of 15 particulate ammonium (NH + 4 ) to NH x dry deposition is considered minor compared to NH 3 (e.g. Asman et al., 1998;Duyzer, 1994). FRAME simulations were combined with land cover data of 25 m×25 m resolution in order to apply land cover specific deposition rates to different land cover types, as described in detail by Vogt et al. (2012b). For the atmospheric inputs of NH x wet deposition and dry and wet deposition of NO y , 20 national modelling at a relatively fine scale resolution, applied to local land cover data, is considered to deliver adequate deposition estimates for this purpose with a relatively low uncertainty in the range of ±20 %.

Agricultural land surface input
Agricultural inputs to the land surface through applications of synthetic fertiliser 25 (N syn fert ), organic fertiliser (N org fert ) and excreta of grazing livestock (N excreta ) were derived from farm activity data (Vogt et al., 2012a). A typical N content was used for the different manure types (Defra, 2010) using grazing records and daily N excretion data as used in the UK NH 3 inventory (Misselbrook et al., 2009). Nitrogen inputs from applications of synthetic fertiliser are considered accurate as this value is known by individual farmers (estimated uncertainty ±10 %). A higher uncertainty of ±30 % is associated with the N input through applications of organic fertiliser, as a typical N content was applied to different manure types 5 as specified by the farmer. The uncertainty associated with the N input through grazing livestock excreta is estimated to be ±50 % as the N content of the grazed grass is not known.

Biological N 2 fixation
Experimentally derived data on biological N 2 fixation are rare in the literature. DeLuca Waughman and Bellamy (1980) measured a fixation rate of 0.7 kg N ha −1 yr −1 in German bogs. The catchment N input through biological N 2 fixation (N biofix ) was thus es-15 timated to be 1 kg N ha −1 yr −1 for both catchments as there was little or no clover in most of the grassland. The N input through biological N 2 fixation is highly uncertain (−70/+300 %) as this term is estimated from only a few experimentally derived literature values.

Gaseous emissions from land surfaces
Ammonia emissions were calculated by applying UK average emission factors (EFs) of the UK emission inventory to the land surface inputs from synthetic and organic fertiliser and grazing excreta (Misselbrook et al., 2009 as discussed in Sect. 2.2. As calculations of NH 3 emissions are based on the local farm inventory and national emission factors, the uncertainty is estimated to be relatively low (±20 %). Direct N 2 O emissions are associated with soil N input (N NH 3 dry dep + N NH x wet dep + N NO y dep +N syn fert +N org fert +N excreta ) and were calculated using the method of Lesschen Emissions of NO were derived by applying a Tier 1 EF of 2.6 % to synthetic fertiliser N applied as recommended in the EEA/EMEP guidelines (McGlade and Vidic, 2009). As there is no specific EF recommended for applications of organic fertiliser and grazing livestock excreta a literature value of 0.5 % was applied (Bouwman et al., 2002). 20 The uncertainty of N 2 O and NO emissions is estimated at ±50 % as they are based on data from the farm inventory and also literature emission factors. Emissions are known to vary substantially depending on soil conditions. Emission factors of N 2 are highly uncertain. Recently, Ammann et al. (2009) applied a literature-derived EF of 12.5 % to N inputs from fertilisation and biological N 2 fixation 25 for a Swiss grassland with an error of ±100 %. For a grazed grassland in southeast Scotland (<10 km from this study landscape), N 2 emissions were modelled and an EF of 10 % of applied N through grazing excreta and synthetic and organic fertilisation calculated (Skiba, personal communication, 2011). This N 2 EF was applied to all fields with agricultural activities in our study catchments. It is noted that there is a large uncertainty (−50/+200 %) associated with this budget term (Sect. 3.7).

Harvested vegetation
Nitrogen output also occurs via removal of vegetation by harvesting (N harvest ) and by grazing livestock (N grass ). The amount of harvested crop and grass removed by farm-5 ers for silage and hay production was derived from the farm survey activity data with a specific N content applied to each main crop type (Møller et al., 2005). The uncertainty of N harvest is thus estimated at ±20 %. The amount of N removed through grass consumption by grazing livestock (N grass ) was estimated as follows: where N excreta is the amount of N excreted by grazing livestock (Sect. 2.3.2), N animal is the N content in the exported wool and meat, calculated using N content values in Roche (1995) and Flindt (2003) and N feed is the N content of the supplementary animal feed, derived by farm activity data and a specific N content of different feed types 15 (Møller et al., 2005). Both N animal and N feed are estimated to have an uncertainty of ±20 %, however considering the ±50 % uncertainty associated with N excreta , the uncertainty of N grass is estimated at ±50 %.

Fluvial export
Annual downstream fluxes (N stream ) of total dissolved nitrogen (TDN), which is the sum 20 of ammonium (NH + 4 -N), nitrate (NO − 3 -N) and dissolved organic nitrogen (DON), were established by Vogt et al. (2012a) by sampling at gauged outlets of the two catchments at both fortnightly and hourly intervals during selected high flow events through 2008. As N stream is based on local measurements conducted throughout the study year, it is considered to carry a relatively low uncertainty, conservatively estimated at ±20 %. Additional information on sources of streamwater N concentrations within the catchments was derived by spatial sampling at stable low flow conditions, conducted in July, September and December 2008.

Results and discussion
The outcomes are explored here using spatially differentiated results of the agricultural 5 land surface N input, the associated land surface N emissions and atmospheric N deposition and fluvial N export. In addition, the catchment N inputs and output terms are summarised and the overall catchment N budgets are given with a discussion of uncertainty. it is noted that there are significant uncertainties associated with the calculation of these N inputs.

Atmospheric N emissions
Gaseous NH 3 emissions from the catchment land surface (excluding housing and manure store emissions) are shown in Fig. 2a. In the Moorland catchment, field based 5 emissions ranged from 0 to 48 kg N ha −1 yr −1 (mean: 0.9 kg N ha −1 yr −1 ) with 58 % originating from applications of organic fertiliser, 40 % from grazing excreta and 2 % from synthetic fertiliser. In the Grassland catchment, NH 3 emissions ranged from 0 to 53 kg N ha −1 yr −1 between individual fields (mean: 4.5 kg N ha −1 yr −1 ) with 66 % arising from organic fertiliser, 30 % from grazing excreta and 4 % from synthetic fertiliser.
Despite most of the agricultural land surface input originating from grazing excreta (Sect. 3.1), the dominant source of NH 3 emissions were applications of organic fertiliser in both catchments, due to high NH 3 volatilisation losses. In contrast, almost all N in grazing excreta (∼95 %) can be expected to enter the catchment soils and thus contribute to soil emissions of N 2 O and N 2 or can be leached. Overall, 7 % of the agri- 15 cultural land surface input of N to the Moorland catchment was estimated to be emitted as NH 3 compared with 9 % from the Grassland catchment. Direct N 2 O emissions from the Moorland catchment averaged to 0.8 kg N ha −1 yr −1 with field emissions ranging from 0 to 7.0 kg N ha −1 yr −1 (Fig. 2b) and from 0 to 36.2 kg N ha −1 yr −1 in the Grassland catchment. However, the uncertainties within those field based emission estimates were relatively large (Table 4) as there is substantial within field variation of N 2 O and N 2 emissions due to the heterogeneity of soil processes (e.g. Hofstra and Bouwman, 2005). Soil NO emissions were estimated to be insignificant for both catchments with emis-5 sions of 0.1 kg N ha −1 yr −1 in the Moorland and of 0.3 kg N ha −1 yr −1 in the Grassland catchment. The field with the highest NO emission was common to both catchments, thus the field specific emission range of 0 to 1.8 kg N ha −1 yr −1 was the same for both catchments.

10
The total atmospheric N deposition to the two studied catchments was estimated to be 8.2 kg N ha −1 yr −1 in the Moorland and 12.3 kg N ha −1 yr −1 in the Grassland catchment (Fig. 3)

Fluvial N export
Both catchments were characterised by highly variable stream flow with high discharge events making an important contribution to annual downstream fluxes (Vogt et al., 2012a). For example, in 2008, the highest 10 % of the discharge data contributed 53 % to the total discharge in the Moorland and 40 % in the Grassland catchment. The an- Maps of annual mean concentrations of NO − 3 , NH + 4 and DON measured during the three spatial sampling campaigns are shown in Fig. 4, together with the underlying land cover. The streamwater NO − 3 concentrations of both catchments have been shown to be significantly positively related to N input through agricultural land surface 15 and atmospheric deposition (Vogt et al., 2012a). Ammonium concentrations were significantly negatively related to N input and could be related to the coverage of wet peaty soils (Vogt et al., 2012a). However, local point source contributions, such as suspected sewage discharge observed while collecting samples, may also contribute to the large spatial variability of NH + 4 concentrations within the Grassland catchment. 20 The sources of DON can vary widely and differed between the catchments (Vogt et al., 2012a). In both catchments, flushing of organic-rich soil water contributed to streamwater DON concentrations, however in the Grassland catchment, there were additional major sources, such as agricultural runoff.
To analyse the potential contribution of the peat cutting area to the DON as well 25 as to the linked dissolved organic carbon (DOC) export flux of the Moorland catchment, the catchment was divided into eight subcatchments based on the drainage pattern. A regression analysis between the % area of peat soil in these subcatchments and DON and DOC concentrations at the subcatchmet outlets mostly showed a positive relationship between DOC and DON concentrations and the % area of peat soil (Fig. 5a, b). This relationship was more pronounced for DOC than DON, however, in both cases there was substantial scatter in the relationship. Other studies (e.g. Aitkenhead et al., 1999) have shown that the area of peat soil in a catchment is directly related 5 to streamwater DOC concentration. Clark et al. (2004) found DON concentrations to be positively related to peat cover in the summer only. In this study, the relationship between DON concentrations and % area of peat soil was also strongest in July. The same regression analysis with % peat cutting area also showed a similar positive relationship to DOC and DON concentrations (Fig. 5c, d) with a slightly stronger relationship observed between concentrations and % peat cutting area (compared to % peat area). This is likely to be a reflection of peat cutting taking place in the areas of deepest peat in the catchment leading to the enhanced effect shown in Fig. 5c "peat effect" on DOC and DON concentrations and contribute to higher annual fluxes because of greater runoff due to drainage. The longer term effect of peat cutting on the catchment fluvial N flux remain a question for further study.

N inputs to land in the study catchments
The various components which contribute N inputs to the two study catchments 25 are summarised in Fig. 6 (input estimates expressed per hectare) and agricultural land surface inputs and 5 % from estimated biological N 2 fixation. Grazing livestock excreta represented the largest single input source, contributing 41 % to the inputs in the Moorland and 40 % in the Grassland catchment. The fraction of the grazing excreta subject to gaseous emissions to the atmosphere (Sect. 3.2) was estimated to be around 21 %, thus the majority of the catchment input through grazing excreta stayed either within the system, i.e. in soil or vegetation, or was leached into surface or groundwaters.

N outputs from land in the study catchments
Catchment outputs are shown as per hectare values in Fig. 7 and as per catchment values in 96 % of N grass in the Grassland catchment. Thus, the main importance of this "grazing livestock N cycle" are increased rates of soil N cycling associated with the grazing excreta which lead to gaseous and streamwater losses. When considering the grazed grass as a recycling budget term, the largest output fluxes of both catchments were the stream exports.

Total N budgets for the study catchments
The overall nitrogen budgets for two catchments are compared in Table 4 and Fig. 8. The Moorland catchment showed a negative N balance of −1.6 + 3.8/−3.4 (error) kg N ha −1 yr −1 , potentially indicating a small loss of N from catchment storage to the stream, however within the uncertainty estimates the catchment N budget could also 10 be in balance. Reynolds and Edwards (1995) , 2004). The differing C balances reflect large inter-annual variability in flux terms, particularly C uptake from the atmosphere which in turn is influenced by the annual fluctuations in weather. Thus, the studied Moorland catchment may shift at an annual level from acting as a net C sink to a source, while at the same time releasing a significant amount of C from the catchment via downstream DOC export. The effects of 5 future climate change on catchment scale C and N budgets remain highly uncertain. The Grassland catchment had a positive N balance of 5.9 + 7.4/−12.3 (error) kg N ha −1 yr −1 , indicating that the catchment stored N inputs in soil, vegetation and groundwater for this study year. However, as with the Moorland catchment, the error bars overlap the balance point. The stream export of the Grassland catchment repre-10 sented a relatively large budget term compared with the other terms. By comparison with other European regional catchment budgets reported by Billen et al. (2011), the retention of N was low (Sect. 4).

Uncertainties in the catchment nitrogen budgets
For both catchments, the budget terms with the largest error bars were the outputs 15 through grazed grass (N grass ) and the input through grazing excreta (N excreta ), as noted above. However, as those terms are interdependent and it is the difference between them that contributes to the overall uncertainty of the N balance calculation, the net error is smaller than the individual errors. In the Moorland catchment, the budget terms contributing the most to the uncertainty of the N balance were biological N 2 fixation, 20 stream export and N 2 emissions. In the Grassland catchment, the most important terms contributing to uncertainty were N 2 emissions, followed by applied organic fertiliser and stream export. The overall uncertainty of the N balances were large, the Moorland catchment balance being −1.6 kg N ha −1 yr −1 with estimated upper and lower balance values of +2.2 and −5.0 kg N ha −1 yr −1 , accounting for uncertainties. Similarly, the up- 25 per and lower estimates of the Grassland catchment of +5.9 kg N ha −1 yr −1 range between +13.3 and −6.4 kg N ha −1 yr −1 . Hence, although we present a detailed budget analysis, the uncertainties remain inherently large. There are several terms still missing from the N budget calculation, which may add further uncertainty to the current balance estimate. In particular, atmospheric deposition of gaseous and particulate organic N compounds were not quantified nor estimated due to lack of information, although organic deposition may be an important input (Cape et al., 2004;Neff et al., 2002). Moreover, fluvial N export through particulate organic 5 N (PON) was not measured, although the PON flux is likely to be insignificant compared to the DON flux as was the POC flux to the DOC flux measured in the Moorland catchment by Dinsmore et al. (2010).
Although our study was detailed, it was carried out over a relatively short time period (one year), which may affect some of the conclusions drawn from the data. In partic-10 ular, stream export fluxes are known to vary year-on-year due to climatic fluctuations (Gascuel-Odoux et al., 2010). Further study on the N budgets of these catchments is needed to clarify the role of annual variation. Another source of uncertainty is the assumption that land use and N input remain approximately constant with time allowing the balancing of N exported through the aqueous system with the N exchange at the 15 surface.

Comparison with a regional catchment N budget approach
Regional scale catchment N budgets have been estimated for many European catchments (Billen et al., 2011). The approach combines a calculation of the net anthropogenic input of reactive nitrogen (NANI, Howarth et al., 1996) to the catchment with 20 data on atmospheric NO y deposition, crop N fixation, fertiliser use and import of food and feed. This is a simple approach which can be applied to large regions, but does not account for processes like NH 3 volatilisation or soil denitrification. In European regional catchments, NANI ranges between 0 and 84 kg N ha −1 yr −1 (mean: 37 kg N ha −1 yr −1 ) (Billen et al., 2011). The relative difference of NANI to the stream export of total N 25 (TN = DIN+DON+PON) is then associated with catchment N retention. Catchment retention refers to the amount of N which is either stored in soils and groundwater or lost through emissions to the atmosphere. There is some evidence that the fraction of NANI exported by the stream is larger in northern European catchments with high discharges. These regional budget calculations differ substantially to the one presented here, (e.g. coarser scale data, no NH x deposition, no land emissions, no organic fertiliser 5 applications); however, the catchment retention calculated as the percentage of the net anthropogenic input which is stored or emitted using our budget terms for the landscape scale may emphasise the differences of regional and landscape scale N budgets. Thus, a landscape NANI was calculated (see Sect. 2.2 for budget term definitions): landscape NANI = N NH 3 dry dep + N NH x wet dep + N NO y dep + N syn fert + N org fert + N excreta The landscape NANI differs to the budget calculation of Eq.
(1) in that biological N 2 fixation, the land emissions and stream export are not taken into account. soils. Expressed in terms of our comprehensive landscape N budgets, the actual "net nitrogen retention" ( [all N input − all N output ]/all N input · 100) would be +9 % and −7.6 % for the Grassland and Moorland catchments, respectively.

Conclusions
Nitrogen budgets for two adjacent catchments with contrasting land use within a single 5 landscape unit were calculated taking into account all agricultural activity and each of the important gaseous and aqueous inputs and outputs. This allowed a detailed analysis of catchment inputs and outputs at a much higher spatial resolution than before.
Within the errors associated with components of the N budget, the two catchments are in an approximate net N balance, although the best estimates suggested a tendency 10 for the Grassland catchment to gain nitrogen (+6 [−6, +13] kg N ha −1 yr −1 ) and for the Moorland catchment to lose nitrogen (−2 [−5, +2] kg N ha −1 yr −1 ). The key uncertainties of our N budget approach were N 2 emissions and stream N export. This emphasises, firstly the need for more studies addressing the quantification of N 2 emissions and, secondly the importance of estimating downstream fluxes accurately. 15 The N budgets of the two study catchments indicate that both catchments have a limited capacity to store nitrogen within soils, vegetation and groundwater. This important finding contrasts with regional scale estimates. The "catchment retention" of N, calculated as the percentage of net anthropogenic N input which is not lost in streamwater (i.e. stored within the catchment or emitted to the atmosphere), amounted to 3 % 20 in the Moorland and 55 % in the Grassland catchment. These values are relatively small compared with estimated catchment retentions in European catchments at the regional scale, ranging from 50 % to 90 % (Billen et al., 2011). Whereas larger, regional scale approaches to estimating catchment input/output may be important for a global overview, these approaches tend to hide the landscape scale N dynamics and thus the This work on compiling landscape scale nitrogen budgets represents the beginning of a better understanding of the anthropogenic impact via agricultural activities on European landscapes. Within the NitroEurope Integrated Project (Sutton et al., 2007), the outcomes of this study are being further analysed in the context of nitrogen fluxes and budgets quantified in different landscapes across Europe, with differing agricultural tre for Ecology & Hydrology, the Scottish Agricultural College, together with complementary inputs from the UK Department of Food and Rural Affairs (Defra), COST 729 and the NinE network of the European Science Foundation. The authors are grateful for the cooperation of all farmers in the study landscape, in particular the poultry farm, for detailed management data. Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Ullyett, J., Bull, K. R., Emmett, B. A., Lowe, J., and Wyers, G. P.: Dispersion, deposition and impacts of atmospheric ammonia: quantifying local budgets and spatial variability, Environ. Pollut., 102, 349-361, 1998. Sutton, M. A., Nemitz, E., Erisman, J. W., Beier, C., Bahl, K. B., Cellier, P., de Vries, W., Cotrufo, F., Skiba, U., Di Marco, C., Jones, S., Laville, P., Soussana, J. F., Loubet,B.,5 Twigg, M., Famulari, D., Whitehead, J., Gallagher, M. W., Neftel, A., Flechard, C. R., Herrmann, B., Calanca, P. L., Schjoerring, J. K., Daemmgen, U., Horvath, L., Tang, Y. S., Emmett, B. A., Tietema, A., Penuelas, J., Kesik, M., Brueggemann, N., Pilegaard, K., Vesala, T., Campbell, C. L., Olesen, J. E., Dragosits, U., Theobald, M. R., Levy, P., Mobbs, D. C., Milne, R., Viovy, N., Vuichard, N., Smith, J. U., Smith, P., Bergamaschi, P., Fowler, D., and