Ammonia deposition in the neighbourhood of an intensive cattle feedlot in Victoria, Australia

Intensive cattle feedlots are large emission sources of ammonia (NH3), but NH3 deposition to the landscape downwind of feedlots is not well understood. We conducted the first study in Australia to measure NH3 dry deposition within 1 km of a commercial beef cattle feedlot in Victoria. NH3 concentrations and deposition fluxes decreased exponentially with distance away from the feedlot. The mean NH3 concentrations decreased from 419 μg N m−3 at 50 m to 36 μg N m−3 at 1 km, while the mean NH3 dry deposition fluxes decreased from 2.38 μg N m−2 s−1 at 50 m to 0.20 μg N m−2 s−1 at 1 km downwind from the feedlot. These results extrapolate to NH3 deposition of 53.9 tonne N yr−1 in the area within 1 km from the feedlot, or 67.5 kg N ha−1 yr−1 as an area-weighted mean, accounting for 8.1% of the annual NH3-N emissions from the feedlot. Thus NH3 deposition around feedlots is a significant nitrogen input for surrounding ecosystems. Researches need be conducted to evaluate the impacts of NH3 deposition on the surrounding natural or semi-naturals ecosystems and to reduce N fertilizer application rate for the surrounding crops by considering nitrogen input from NH3 deposition.


Results
The dynamics of hourly air temperature, relative humidity and wind speed during the three sampling periods in April, May and June are shown in Fig. 1. The mean daily maximum/minimum temperatures were 21.9/6.4 °C, 19.8/3.5 °C, 13.5/5.3 °C respectively, during the sampling periods of April, May and June, which were 4 °C, on average, lower than the corresponding annual means of the daily maximum/minimum temperatures for the year 2015 19 . The mean wind speeds during the three sampling periods were 4.3, 2.6 and 1.9 m s −1 respectively. The averaged wind speed during the three periods (3.0 m s −1 ) was comparable to the annual mean wind speed in 2015 (3.0 m s −1 ) 19 . The mean values of relative humidity during the three periods were 61.1, 81.1 and 71.8% respectively, with a mean of 71%, which was approximately one-third higher than the annual mean in 2015 19 .
The measured NH 3 concentrations showed large spatial and temporal variation (Fig. 2). During the three sampling periods, the mean daytime and nighttime (daytime/nighttime) NH 3 concentrations were 300/370, 217/324, 117/245, 61.5/181 and 27.2/94.2 μ g N m −3 at the distance of 50, 100, 200, 500 and 1000 m downwind from the feedlot, respectively. From 50 m downwind to 1 km downwind, NH 3 concentrations decreased by 74 to 97% during the daytime, and 60 to 87% during the nighttime. NH 3 concentrations decreased exponentially with distance away from the feedlot (Fig. 3). This indicates, during the period of transporting NH 3 from the feedlot to the sampling sites, that NH 3 deposition and NH 3 dispersion might occur 20 . NH 3 concentrations were observed the highest at night under stable atmospheric conditions when low dispersion of the NH 3 plume from the feedlot occurred. The average NH 3 concentrations during the nighttime were 1.1-6.0 times of those during the daytime.  The ratio of NH 3 concentration during the nighttime to that during the daytime increased with the distance away from the feedlot, due to the faster decrease of NH 3 concentration with distance away from the feedlot during the daytime than during the nighttime (Fig. 3). This suggests that sampling NH 3 concentrations separately during the daytime and nighttime is required.
The NH 3 deposition fluxes also showed large spatial and temporal variations (Fig. 4), ranging from 0.05 to 2.94, 0.03 to 4.34 and 0.03 to 4.34 μ g N m −2 s −1 in April, May and June respectively for the cropland. The NH 3 deposition fluxes decreased with the distance away from the feedlot. Higher NH 3 deposition fluxes were generally found during the daytime than during the nighttime at the 50 m site, but this trend was reversed at the 200 m to 1000 m sites. This may be because that the difference of NH 3 concentration between daytime and nighttime was relatively small at the 50 m site (Fig. 4) and therefore higher wind speed during the daytime favored lower aerodynamic resistance (R a ) and caused higher deposition flux in the daytime 14 . But the much higher NH 3 concentration during the nighttime than during the daytime at 200 m, 500 m and 1000 m sites made NH 3 concentration the major factor affecting NH 3 deposition flux.
The mean NH 3 fluxes in the three sampling periods in April, May and June under the land use types of cropland, grassland and pasture are summarized in Table 1. Similar to NH 3 concentration, the mean NH 3 flux also decreased exponentially with the distance away from the feedlot. The variation of the mean NH 3 deposition fluxes among the three land use types was small. The similarity may be ascribed to the small differences of input parameters (e.g. R a , quasi laminar boundary layer resistance (R b ), stomatal resistance (R s ), cuticular resistance (R w )) that were used for calculating NH 3 bi-directional exchanges for the three land use types [21][22][23] (see the Supplementary  Information for more details).
We estimated the annual NH 3 dry deposition rates at the downwind sites by assuming that the mean NH 3 deposition fluxes during April -June for cropland, grassland and pasture represented the annual average of NH 3 deposition fluxes. The estimated annual NH 3 deposition rates were 614, 496, 322, 210 and 106 kg N ha −1 yr −1 at the downwind sites with 50, 100, 200, 500 and 1000 m respectively, from the feedlot, provided that wind direction was constant. In fact, NH 3 deposition mostly occurred in the downwind areas of the feedlot in this study since the NH 3 concentration was very low or could not be detected in the upwind direction. Due to the frequent changes in wind direction in the studied region, we calculated the annual NH 3 deposition in the downwind areas of eight major wind directions (Table 2) by integrated the site-specific NH 3 dry deposition rates in the downwind transect. By summation, the estimated total NH 3 -N deposition in the areas within 1 km from the feedlot was 53.9 tonne yr −1 , or 67.5 kg N ha −1 yr −1 as an area-weighted mean.

Discussion
The estimated total annual NH 3 deposition in the area within 1 km away from the studied feedlot accounted for 8.1% of the annual NH 3 emissions from the feedlot (664 tonne NH 3 -N yr −1 ). This was comparable to that reported by Fowler et al. 13 , who estimated that 5-10% of NH 3 emissions were dry deposited within 1 km of a poultry farm. However, some other studies reported higher fractions of emissions deposited locally. For example, Hao et al. 24 estimated that 16% of the total NH 3 emitted from a 25,000-head cattle feedlot was deposited to the soil within 1 km of the feedlot. Similarly, Walker et al. 14 calculated that 10.4% of the emitted NH 3 was dry deposited within 500 m of the emission source. Modelling results have shown that the fraction of local deposited NH 3 emissions ranged from 2% to 55% within 1 km of the source, which mainly depending on source height, wind speed,   atmospheric stability, structure of the surrounding canopies and surface resistance 20 , though most estimates are generally smaller than 20% 1,2 . One possible reason for the smaller fraction of emissions deposited locally in this study could be attributed to the relatively high wind speed at our site (with an annual mean of 3 m s −1 at 2 m height). High wind speed usually favors a high NH 3 emission rate 18 , but may also cause fast dispersion and dilution of the NH 3 plume and thus cause low NH 3 concentration as well as low NH 3 dry deposition in the downwind areas 20 . The reduced NH 3 deposition in the downwind areas and the increased NH 3 emission rate from feedlot due to high wind speed then may have resulted in a relatively low fraction of locally deposited emissions. For the remaining 92% of the NH 3 emitted from the studied feedlot, one possible fate may be that it was transported to the mixing layer in the downwind regions of the feedlot by turbulent dispersion and advection. The depth of the mixing layer has been reported to range from 100 m to up to 1500 m 25,26 . Usually, the larger the depth of the mixing layer, the more favourable the mixing of air pollutants with elevation 25,26 . In May of 2015 an investigation of NH 3 concentration around the same feedlot as this study, using the airborne technology, found NH 3 concentration to range from 470 ppb (294 μ g N m −3 ) at 0.5 km downwind of the feedlot to 25 ppb (16 μ g N m −3 ) at 6.5 km downwind of the feedlot at 35 m above ground, and as high as 40 ppb (25 μ g N m −3 ) at 310 m above the ground along the boundary of the feedlot 27 . These results indicate that most of the emitted NH 3 may be transported to the mixing layer and could be transported at least 6.5 km from the feedlot. Due to the small emission intensities of acidic gases (e.g., nitric oxides, sulphur dioxide) from industrial and transport sources and nearly no aerosol pollution in the region of the feedlot, the transformation of NH 3 to particulate NH 4 + could be ignored in the neighbourhood of the feedlot. Therefore, it is also worthwhile to further investigate NH 3 deposition and related environmental effects in the downwind areas 6.5 km or more from intensive feedlots.
The measured NH 3 concentrations and estimated NH 3 deposition rates 1 km from the cattle feedlot in this study were higher than those reported from poultry farm or swine production facilities 13,14 . For example, the annual mean NH 3 concentration was 19 to 52 μ g N m −3 at a distance of 15 m from a poultry farm (emission intensity: 4.8 tonne NH 3 -N yr −1 ) in the UK and declined to background concentrations of 0.8 to 1.6 μ g N m −3 at a distance of 270 m, while NH 3 deposition decreased from 42 kg N ha −1 yr −1 at 15 m to 5 kg N ha −1 yr −1 at 270 m with an average of 7 kg N ha −1 yr −1 in the area within 300 m of the poultry farm 13 . Furthermore, the measured NH 3 concentration ranged from 139 μ g N m −3 at a distance of 10 m from a swine production facility in eastern North Carolina (emission intensity: 28.2 tonne NH 3 -N yr −1 ) to 10.7 μ g N m −3 at 698 m, while the mean NH 3 deposition ranged from 26 to 52 kg N ha −1 yr −1 within 500 m from the source 14 . As the estimated NH 3 emission intensity of the source in this study (664 tonne NH 3 -N yr −1 based on 17,500 head of cattle and an emission factor of 104 g NH 3 -N head −1 d −1 ) was also higher than those in the above two studies, it can be concluded that the sources with high NH 3 emission intensities will lead to high NH 3 deposition in the surrounding area. This conclusion is also supported by Hao et al. 24 , who reported much higher NH 3 deposition around a 25,000-head beef feedlot (emission intensity: 235 tonne NH 3 -N yr −1 ) in Alberta, Canada, which ranged from 120 kg N ha −1 yr −1 50 m from the source to 20.8 kg N ha −1 yr −1 1 km from the feedlot.
It should be noted that the study we report here is a pilot study to apply the bi-directional NH 3 exchange model to estimate NH 3 deposition near a feedlot in Australia. Our calculated NH 3 deposition is still subject to uncertainty in the model input parameters (R a , R b , R s , R w , R g , χ s and χ g , see the Supplementary Information for the definitions of these parameters) due to that parameterization of these variables in our study were mainly using the equations or empirical values based American or European researches. For evaluation of the whole model, we also calculated NH 3 dry deposition velocity by dividing the NH 3 deposition flux by NH 3 concentration since no NH 3 emission flux occurred in this study. The NH 3 deposition velocities were on average 0.5-0.6 cm s −1 for cropland, pasture and grassland around the feedlot. These deposition velocities are comparable with those published mean NH 3 deposition velocities for cropland (0.4-0.8 cm s −1 ) 28,29 , pasture (~0.8 cm s −1 ) 30 and grassland surfaces (0.5-1.0 cm s −1 ) 28 . Therefore, though there are some uncertainties, the calculated NH 3 deposition fluxes in this study are still in a reasonable range. Due to the difficulty in accessing the cropland to install the NH 3 samplers in the growing season, we only conducted NH 3 deposition sampling during three months. Underestimation may exist when using these three months data to estimate NH 3 deposition for the whole year. Firstly, due to that the mean air temperature during the three sampling periods in the study was 4 °C lower than the annual mean of air temperature and NH 3 emission rate is positively correlated with air temperature 18 , the NH 3 emission intensity and thus the measured NH 3 concentration during the sampling periods may be lower than the annual means, which means the annual NH 3 deposition rate at the sampling sites might be underestimated. Secondly, during the three months of sampling periods, only one month was classified into growing season and the other two months classified into un-growing season. In fact, there are half growing season and half un-growing season in a year in the studied region. Because the R s and R w are smaller in growing season than those in un-growing season, which favors faster deposition of NH 3 in growing season than those in un-growing season, the reduced duration of growing season (by 17%) in a year may cause another underestimate of the annual NH 3 deposition rates.
Our study and previous investigations all indicated that there was high NH 3 deposition around intensive feedlots or animal farms, which was ranged from 20 to 120 kg N ha −1 yr −1 and usually higher than the critical loads (10-15 kg N ha −1 yr −1 ) of N deposition for most natural or semi-natural ecosystems 31 . Former studies had documented that loss of biodiversity, soil acidification, increase of soil N 2 O emissions could occur in natural and semi-natural ecosystems with increased atmospheric N deposition [6][7][8]32 . Therefore, considering the high NH 3 deposition around the feedlots or animal facilities, comprehensive studies should be conducted to evaluate the impacts of NH 3 deposition on the surrounding natural or semi-natural ecosystems, especially in those regions already suffering from high background N deposition (e.g. Eastern China 33 , Western Europe 28 ). For the croplands or pastures around the feedlots or animal facilities, NH 3 deposition is an important N source and researches also need be conducted to reduce N fertilizer application rate by considering N input from NH 3 deposition in fertilizer recommendation so as to avoid excessive N fertilizer application.

Methods
The experiment was conducted at a typical intensive cattle feedlot, in northeastern Victoria, Australia (Fig. 5). The region has a Mediterranean climate, with long hot summers and mild wet winters 19 . From 2004 to 2015 the mean maximum/minimum temperature was 23.1/9.0 °C and mean annual precipitation was 355 mm 19 . The feedlot area was approximately 93 ha (1,230 m in the east-west direction and 760 m in the north-south direction) including cattle pens, manure stockpiles, bare soil or roads and effluent ponds. The feedlot held approximately 17,500 cattle during the study period. The cattle were 1-1.5 yr of age, European breeds, mostly Angus and Angus cross, with an average body weight of 396 ± 5.3 kg. The cattle consumed an average of 10.2 kg dry matter daily of a finishing ration of barley (Hordeum vulgare L.) and grass hay 18 . The area surrounding the feedlot was mainly cropland planted to wheat with sheep pasture lying to the northeast and grassland to the south (Fig. 5). The wheat croplands around the feedlot were usually fertilized (urea was used as basal nitrogen fertilizer at an application rate of approximately 50 kg ha −1 ) and sown in June and harvested in October or November, and fallowed from November or December until May in the following year. The plants in the pasture and grassland were usually growing during June to September.
The NH 3 deposition measurement was conducted during three sampling periods in 2015 (20 th to 24 th of April, 14 th to 18 th of May and 24 th to 29 th of June). The NH 3 concentration was measured by a denuder system for long-term ammonia sampling (DELTA) [34][35][36] . A low-volume pump (D210, TCS Micropumps Ltd., UK) was used to draw air at a rate of 0.2-0.4 L min −1 . Prior to the measurement, two denuders, connected in series to adsorb the NH 3 in the air, were treated with a solution of 5% (m/v) citric acid in methanol. When the air was drawn through the denuder train, NH 3 gas was adsorbed to the inner surface of the denuders. The total sampled air volume was recorded by a dry gas meter (SK25, Kimmon Manufacturing Co., Ltd., Japan). The gas meter was checked at the start of each sampling period using a gas flow meter to ensure that the recorded gas volumes were correct. The denuders, gas meter and pump were connected in sequence with short (2-3 cm) silicone tubes, and were fixed in a PVC box (40 × 30 × 50 cm) with the inlet of the denuder train exposed to the ambient air. During NH 3 sampling period, the PVC box was attached to a pole at a height of 1.5 m above the ground. The denuder trains were changed two times each day, based on the day and night cycle. During each sampling period, five daytime (8:00 am to 5:00 pm) and nighttime (6:00 pm to 7:00 am in the following day) NH 3 samples were collected continuously at each location. The samples were stored at 4 °C and analysed at the end of each sampling period at an off-site laboratory. The NH 3 denuders were extracted with milliQ ® water for 1 h, and the extraction was analysed for NH 4 + -N content by segmented flow analyzer (Skalar SAN ++ , Netherlands). The detection limit of the DELTA system, calculated as 2σ (two times of the standard deviation) of the field blanks, was 0.28 μ g N m −3 .
The NH 3 concentrations were measured at five locations downwind of the feedlot. The sampling locations were selected along a transect downwind of the centre of the feedlot, with a distance of 50, 100, 200, 500 and 1000 m from the fence line of the feedlot (Fig. 5). The sampling locations were determined according to daily predominant wind direction, measured by a three-dimensional (3-D) sonic anemometer, so that downwind NH 3 concentrations from feedlot were measured. The actual sampling duration for a sampling site was then recorded only when the site was located at downwind of the feedlot. The measured NH 3 concentrations were discarded if wind direction changed and the downwind sampling duration was less than 50% of the total sampling duration.
A weather station coupled with a 3-D sonic anemometer (CSAT3, Campbell Scientific, Logan, USA) was set up at a height of 3.3 m above the ground located to the east of the feedlot. Fifteen-min averaged air temperature, wind speed, friction velocity, Monin-Obuhkov length and relative humidity were recorded at 10 Hz. The raw data was processed to hourly average data using SAS software (SAS 9.4, SAS Institute Inc., Cary, NC, USA).
A well tested bi-directional NH 3 exchange model, which is called the two-layer canopy compensation point model 23,37 , was used to estimate NH 3 dry deposition around the feedlot. Similar to most of the dry deposition models, the bi-directional flux model is based on a formula analogous to Ohm's law in electrical circuits where flux (analogous to current) is calculated by dividing the concentration difference (analogous to voltage) by the deposition resistance (analogous to electrical resistance) 21,38 . According to Nemitz et al. 37 , the total NH 3 flux (F t ) is the sum of bi-directional exchange with the leaf stomata (F s ), deposition to the leaf cuticle (F w ) and bi-directional exchange with ground (F g ). Among these pathways, F s and F w occur parallelly in the canopy layer and can be summed as the canopy flux (F f ), while F g occurs in the ground layer 37 . The relations and definition of each of the fluxes are shown in the following equations 37 : where χ a is the measured NH 3 concentration at the height of 1.5 m above the ground level, χ z ( ) 0 is the NH 3 concentration at height of d + z 0 , d is the zero-plane displacement height, z 0 is the surface roughness length, R a is the aerodynamic resistance, a function of the vertical turbulent diffusive transport through the atmosphere, χ c the canopy NH 3 compensation point, R b the quasi laminar boundary layer resistance, χ s the stomatal compensation point, R s the stomatal resistance, R w the cuticular resistance, χ g the ground layer NH 3 compensation point and R g the in-canopy resistance to the ground. Based on the equations of (1), (3), (4) and (7), χ z ( ) 0 can be calculated using the following equation: Based on the equations of (2), (4), (5), (6) and (8), the equation of χ c can be deduced as the following: The parameters of R a , R b , R g , R s , R w , χ g and χ s were calculated according to Wesely 21 , Erisman & Draaijers 22 and Massad et al. 23 (see the Supplementary Information for more details). Therefore, according to equations (3,8 and 9), the total NH 3 dry deposition flux was calculated. Hourly NH 3 dry deposition flux was calculated based on the hourly meteorological data. As we did not measure hourly NH 3 concentrations, their values in a sampling event were assumed to be equal to the corresponding daytime or nighttime NH 3 concentration measured by the DELTA system. The missing data of NH 3 concentrations in a sampling event were interpolated with the mean daytime or nighttime NH 3 concentrations measured during the same sampling period. The R software v3.13 39 was used for regression analyses. The significance level was set as p < 0.05. The following steps were used to integrate the site-specific NH 3 dry deposition rates in the downwind transect to the total NH 3 deposition in the downwind area within 1 km from the feedlot. Firstly, using the correlation between the measured NH 3 deposition rate and distance from the feedlot, we estimated the deposition rates at 200, 300, 400, 600, 700, 800 and 900 m from the feedlot. Secondly, due to that NH 3 deposition occurred mostly in the downwind direction as the NH 3 concentration was very low or could not be detected, we divided the total area within 1 km from the feedlot into eight downwind areas based on the eight major wind directions. Here a downwind area is defined as the area within 1 km from the feedlot that can be affected by the feedlot NH 3 plumes transported by a certain direction of wind. Examples for dividing downwind areas are shown in Fig. 6. We further divided each downwind area into 11 sub-areas, that is 1) area within 50 m from the feedlot, 2) area within 50 to 100 m from the feedlot, 3) area within 100 to 200 m from the feedlot, 4) area within 200 to 300 m from the feedlot, 5) area within 300 to 400 m from the feedlot, 6) area within 400 to 500 m from the feedlot, 7) area within 500 to 600 m from the feedlot, 8) area within 600 to 700 m from the feedlot, 9) area within 700 to 800 m from the feedlot, 10) area within 800 to 900 m from the feedlot, and 11) area within 900 to 1000 m from the feedlot. For the Scientific RepoRts | 6:32793 | DOI: 10.1038/srep32793 sub-area 1), we assumed that the NH 3 deposition rate in this area was equal to the NH 3 deposition rate at 50 m from the feedlot. NH 3 deposition rates in other sub-areas were assumed to be equal to the mean of the deposition rates at the nearest and longest distances from the feedlot in the specified sub-area. For example, for sub-area 2), the NH 3 deposition rate in the area was assumed to be equal to the mean of the deposition rates at 50 m and 100 m from the feedlot. The NH 3 deposition rate in each sub-area then can be calculated by multiplying the site-specific NH 3 deposition rate with the size of each sub-area. Thirdly, we calculated NH 3 deposition in a downwind area by multiplying the frequency of a wind direction in a year with the summed NH 3 deposition in 11 sub-areas of this downwind area. By summation, we can get the total NH 3 deposition within 1 km from the feedlot using the following equation: where T D is the total NH 3 deposition (tonne N yr −1 ) in the area within 1 km from the feedlot, A ij is the size (ha) of the j th sub-area of the i th downwind area, D j the NH 3 deposition rate (kg N ha −1 yr −1 ) in the j th sub-area, f i is the frequency of the i th wind direction in a year and and 1000 the unit conversion factor. The area-weighted NH 3 deposition rate is then calculated by divided the total NH 3 dry deposition around the feedlot by the total area within 1 km from the feedlot.