Environmental and Animal-Related Parameters and the Emissions of Ammonia and Methane from an Open-Sided Free-Stall Barn in Hot Mediterranean Climate: A Preliminary Study

: Increased knowledge on the factors that affect emissions from open-sided dairy buildings may lead to an improvement of the mitigation strategies. In this study, ammonia (NH 3 ) and methane (CH 4 ) emissions were assessed in an open dairy barn in a hot Mediterranean climate at different managements of the cooling system, as well as the inﬂuence of environmental and animal-related parameters on daily emissions. Measurements of gas concentrations and micro-climatic parameters were carried out in a cubicle free-stall dairy barn located in the province of Ragusa (Italy) in two weeks of 2016 characterised by similar climatic conditions in the warm period. Emissions of NH 3 and CH 4 were estimated through the application of the carbon-dioxide (CO 2 ) mass balance method. Data collected were organised in speciﬁc datasets to carry out different statistical analyses on gas emissions depending on selected parameters for the two weeks with a different management of the cooling system. The results showed higher NH 3 emissions and lower CH 4 emissions in W1 than those in W2. The variability in gas emissions was related to the effect of temperature humidity index (THI) ( p < 0.001) and cow behaviour ( p < 0.01). The highest emissions were recorded during the cleaning procedures for both NH 3 ( p < 0.001) and CH 4 ( p < 0.001), whereas the lowest emissions were recorded during the central hours of the day.


Introduction
The emissions of ammonia (NH 3 ) and methane (CH 4 ) from the livestock sector have great environmental impacts. In 2017, manure management from dairy cattle produced 11% of the NH 3 emissions in the EU28 [1], whereas in 2015, the absolute emissions from the dairy cattle sector was 1711.8 million metric tonnes CO 2 equivalent; in particular, about 1100 was for CH 4 [2] derived from enteric fermentation and manure management. Increased knowledge on the factors that affect emissions from dairy buildings may lead to an improvement of the mitigation strategies for emissions reduction.
In Europe, dairy cows are mainly housed in naturally ventilated (NV) barns [3,4]. In these structures, which are characterised by large side openings, outdoor climatic parameters influence the indoor environment due to the high indoor-outdoor air exchange. In fact, the increase in outdoor temperature during spring and summer seasons in this housing system produces the highest emissions in the barn due to the increase in indoor temperature. As hot climate conditions produce heat stress on the cows [5][6][7], the NV barns are equipped with a cooling system (e.g., fans and sprinklers) to mitigate the effect of heat stress on the cows [8,9], especially in hot climate Mediterranean regions.
In the literature, many studies have been performed to assess the influencing parameters on gas emissions. Saha et al. [10] found that wind speed and wind direction, as well as The natural ventilation system was integrated by two cooling systems, composed of a fogging system with fans, located in the resting area, and a sprinkler system with fans, located in the feeding alley. In detail, the fans in the resting area had a width of 1400 mm, tilt angle of 20° from the horizontal and a ventilation rate of 34,600 m 3 /h. The rotation axis was located at 2.75 m from the floor along the longitudinal axis of the barn, and the distance between fans was equal to 14.0 m. The fogging system in the resting area was composed of misters having an operating pressure of 200 kPa and the corresponding rate of 1.01 L/min for each nozzle. They were located at a height of 2.90 m from the floor and spaced about 3.10 m along the longitudinal axis of the barn. The half-circle (180°) sprinklers in the feeding alley had an operating pressure of 200 kPa and a rate of 2.57 L/min. The sprinklers were installed above the rack at a 2.0 m height from the floor along the longitudinal axis of the feeding alley with a 1.90 m spacing. Each axial fan in the feeding alley had a diameter of 900 mm, tilt angle of 20° from the horizontal and a ventilation rate of 22,250 m 3 /h. The five axial fans were located above the feeding alley with the rotation axis at 2.70 m from the floor and parallel to the longitudinal axis of the feeding alley. The distance between fans in this row was 14.0 m.

Measurement of Gas Concentrations and Climatic Parameters
Concentrations of NH3, CH4 and carbon dioxide (CO2) were continuously acquired at five sampling locations (SLs) by an INNOVA photo-acoustic analyser composed of a Multigas Monitor mod 1412i and a multipoint sampler 1409/12 (Lumasense Technology A/S). The five inlet channels were made of PTFE (polytetrafluoroethylene) tubing to reduce the adsorption of samples. The end of each sampling tube had an air filter to keep the 1409 free of particles. The accuracy of the Multi-Gas Analyser involved a 2-3% absolute deviation in concentrations, and the detection limits were 0.2 ppm for NH3, 0.4 ppm for CH4 and 1.5 ppm for CO2. The SLs were located in different areas of the barn. The location of the four indoor SLs was in the central area of the barn at 20 cm from the barn floor. The background concentrations were recorded in an outdoor SL, located at 3 m The natural ventilation system was integrated by two cooling systems, composed of a fogging system with fans, located in the resting area, and a sprinkler system with fans, located in the feeding alley. In detail, the fans in the resting area had a width of 1400 mm, tilt angle of 20 • from the horizontal and a ventilation rate of 34,600 m 3 /h. The rotation axis was located at 2.75 m from the floor along the longitudinal axis of the barn, and the distance between fans was equal to 14.0 m. The fogging system in the resting area was composed of misters having an operating pressure of 200 kPa and the corresponding rate of 1.01 L/min for each nozzle. They were located at a height of 2.90 m from the floor and spaced about 3.10 m along the longitudinal axis of the barn. The half-circle (180 • ) sprinklers in the feeding alley had an operating pressure of 200 kPa and a rate of 2.57 L/min. The sprinklers were installed above the rack at a 2.0 m height from the floor along the longitudinal axis of the feeding alley with a 1.90 m spacing. Each axial fan in the feeding alley had a diameter of 900 mm, tilt angle of 20 • from the horizontal and a ventilation rate of 22,250 m 3 /h. The five axial fans were located above the feeding alley with the rotation axis at 2.70 m from the floor and parallel to the longitudinal axis of the feeding alley. The distance between fans in this row was 14.0 m.

Measurement of Gas Concentrations and Climatic Parameters
Concentrations of NH 3 , CH 4 and carbon dioxide (CO 2 ) were continuously acquired at five sampling locations (SLs) by an INNOVA photo-acoustic analyser composed of a Multigas Monitor mod 1412i and a multipoint sampler 1409/12 (Lumasense Technology A/S). The five inlet channels were made of PTFE (polytetrafluoroethylene) tubing to reduce the adsorption of samples. The end of each sampling tube had an air filter to keep the 1409 free of particles. The accuracy of the Multi-Gas Analyser involved a 2-3% absolute deviation in concentrations, and the detection limits were 0.2 ppm for NH 3 , 0.4 ppm for CH 4   Climate and microclimatic parameters were measured by sensors installed inside the barn at 2.0 m above the floor and outside the barn above the roof along the ridge. Air temperature, relative humidity, indoor air and wind speed and direction were acquired with a frequency of five seconds by a data-logger CR10X (Campbell, UK) that, every five minutes, computed the average values and stored them in memory locations. Air temperature (Rotronic Italy s.r.l., Milano, Italy) was measured by platinum thermo-resistances (Pt 100 Ω 0 • C) characterised by a measurement range between −40 and +60 • C and a precision of ±0.2 • C (at 20 • C), whereas air relative humidity was measured by an hygrometer (Rotronic Italy s.r.l., Milano, Italy) with a sensitivity of ±0.04%RH/ • C and a precision of ±2% (at 20 • C). The anemometers (WindSonic, Gill instruments Ltd., Lymington, UK) with two-dimensional sonic sensors had a velocity measuring interval of 0−60 m s −1 , a precision of ±2% (at 12 m s −1 ), a resolution of 0.01 m s −1 and a threshold of 0.01 m s −1 ; and had a direction measuring interval of 0−359 • , with a precision of ±3% (at 12 m s −1 ) and a resolution of 1 • .

Barn Management and Behavioural Activity Recordings
Sixty-four Fresian cows were housed in the cubicle free-stall barn with an average weight of 643 (±72) kg and milk yield of 37.2 kg/day. During the trial, the diet of the cows, based on a mixed ratio with concentrate, was the same in both weeks. The activation of the sprinklers and fans was not simultaneous and the activation time is reported in Table 1. In detail, the fans were automatically switched off during wetting to avoid the scattering of water. The cooling systems were manually switched off during the milking sessions and the cleaning of the feeding alley. During the two weeks of observation, the management of the cooling system was different. In detail, the sprinklers in the feeding alley were not operated in W2. The daily routine of the cows was further affected by the barn management. In detail, the cleaning of the barn was carried out once a day at about 07:30 a.m. by a mechanical tractor with scraper; the milking session was performed twice a day at about 5:00 a.m. and 5:00 p.m.; the feed was delivered every day after cleaning, and it was moved into the manger before the first and the second milking sessions. Table 1. Timetable of the two cooling systems in the two weeks.

System
Area Activation Time

Fans
Feeding area 8.30-9.00 Sprinkler system with fans Feeding area 9.00-11.00 Fogging system with fans Resting area 11.00-16.00 Sprinkler system with fans Feeding area 16.00-17.00 Sprinkler system with fans Feeding area 18.30-20.00 Fans Resting area 20.00-1.00 A 24 h video-recording system, composed of ten cameras (Kon.Li.Cor, Perugia, Italy), recorded cow behaviour during the experiment in pen 1 and pen 2 ( Figure 1). Based on the visual evaluation of the records, the scan sampling method [22] was applied with a frequency of 15 min in order to assess animal behaviour in pen 1 and pen 2 (i.e., forty cows were observed with cameras). Under this method, a skilled operator counted the number of cows in a specific behaviour (i.e., feeding, lying and standing) in each frame of the video recordings and stored them in a spreadsheet. Then, the CFI, CLI and CSI were computed [19]. In detail, the CLI was determined by counting the number of cows in lateral or sternal recumbence inside the stall and applying the following relation: CLI = (Cows Lying in Stall)/(Total Cows). The CSI, related to the number of cows in upright posture, walking included, was determined by the following equation: CSI = (Cows Standing in Stall)/(Total Cows). The CFI, based on the number of cows that ingest feed or water, was evaluated with the equation: CFI = (Cows Feeding in Stall)/(Total Cows).

Air Exchange Rate and Emissions Calculations via Indirect Method
In this study, the estimation of NH 3 and CH 4 emissions was carried out by applying the CO 2 mass balance method [3,10,23]. The ventilation rate Q (m 3 h −1 ) was calculated as follows: where P CO2 represents the excretion rate of CO 2 from one cow (g cow −1 h −1 ), N is the number of cows housed inside the building, and C CO2in and C CO2out are the hourly average concentrations of the gas inside and outside the building in the prevailing air direction, respectively (g m −3 ). Within the CO 2 balance, the CO 2 excretion rate depends on animal heat production. As, in this barn facility without deep litter, the CO 2 production from manure is below 4% of the total production [24,25], the CO 2 excretion rate was computed as follows [26]: where q t is the total heat production (W), m is the average mass of the animals (kg cow −1 ), p is the number of days after insemination (d), y is the milk yield (kg d −1 ), CF is the temperature correction factor, T i is the temperature inside the building ( • C) and q cor is the corrected value of the total heat production (W). The average weight, milk yield of the cows and contribution of pregnancy were 650 kg cow −1 , 32 kg d −1 and 135 days, respectively. The emission rate of NH 3 and CH 4 was estimated using the following equation: where E t is the emission rate of the gas (g h −1 ), Q is the ventilation rate calculated according to the CO 2 balance method (m 3 h −1 ), and C in and C out are the average concentrations of the gas inside the barn (i.e., average value of the 4 indoor SLs) and outside the building (Figure 1), respectively (g m −3 ). The weight of the cows and the production may differ from herd to herd. In order to make the results comparable with other studies, the emissions Et were computed as the emissions per livestock unit (LU) in g h −1 LU −1 , where LU is equivalent to a 500 kg animal mass. The emission rate per LU is estimated as follows:

Data Analyses
Data were analysed by using Microsoft ® Excel and Minitab software. Statistical analyses were carried out to assess the influence of a different management of the cooling system in the two weeks. For each week, the influence of THI and cow behaviour on daily gas emissions was analysed. The THI was computed according to Hempel et al. [27]: where T is the dry bulb air temperature ( • C) and RH is the relative humidity (%). Gas emissions were subdivided considering the following groups of THI: comfort zone (THI ≤ 72); low risk of thermal stress for cows (72 < THI ≤ 78); thermal stress (78 < THI < 84). As conditions of emergency for cows (THI ≥ 84) were not recorded in this study, emissions for emergency conditions were not analysed.
The influence of cow behaviour on hourly gas emissions was analysed by using the CLI and the cow activity index (CAI), computed as the sum of CFI and CSI. For each value of gas emissions, a specific cow behaviour (i.e., cow activity and cow lying) was associated. The cow lying condition is related to emissions when cows are resting in the cubicles for a mean value of CLI > 0.50 for each hour. The cow activity condition includes gaseous emissions produced when cows are standing, feeding and walking for a mean value of CAI > 0.50 for each hour. As, during the cleaning of the floor, gas concentrations and emissions were found the highest [28,29], an additional group, named barn cleaning, was considered. Barn cleaning refers to the dataset of gas emissions with a mean value of CAI > 0.50 for each hour that is related to the cleaning of the floor (i.e., when cows are either feedingor standing or walking).
The statistical analyses were carried out by using gas emissions in the prevailing air direction from North-West (135-225 • ), which represented about the 85% of the whole dataset. Moreover, the final dataset of gas emissions was composed of the values that excluded gas emissions under complex windy conditions [23], which represented about 1% of the dataset in the prevailing air direction. In detail, CO 2 differences lower than 100 ppm were neglected to reduce biases in the emission values. After checking the assumptions, a one-way analysis of variance (ANOVA) was applied to the above-described groups of NH 3 and CH 4 emissions. The level of significance was defined by a p value of 0.05. If the test result was significant, the Tukey test post hoc analysis was applied in order to identify differences between groups of gas emissions.

Results and Discussion
The results showed a daily variation in NH 3 and CH 4 emissions due to the microclimatic conditions and barn management. The different management of the cooling system in the two weeks under study modified the cows' behaviour in response to heat stress. Consequently, the NH 3 and CH 4 emissions in the barn were significantly different in the two weeks analysed. The choice of the two weeks is related to the identification of two periods with homogeneous microclimatic conditions in the barn. In fact, microclimatic parameters (i.e., air temperature, air velocity and relative humidity) were not significantly different (p < 0.001) in both W1 and W2. In this way, the difference between the two periods was due to the different management of the cooling system. The outcomes of this study constitute preliminary information for the computation of emissions from the open-sided dairy barn, through the increase in knowledge on how gas emissions, through environmental and animal-related parameters, are influenced by different managements of the cooling system.

Daily NH 3 and CH 4 Emissions
The mean NH 3 emission rate in W1 was 4.73 g LU −1 h −1 with an indoor average temperature of 26.3 • C and an air velocity of 0.71 m s −1 , whereas the mean NH 3 emission rate in W2 was 5.90 g LU −1 h −1 with an indoor average temperature of 27.5 • C and an air velocity of 0.69 m s −1 . The hourly average NH 3 emissions were 0.85 to 11.25 g LU −1 h −1 in W1 and 1.64 to 10.86 g LU −1 h −1 in W2. The values of NH 3 emissions found by other studies [3,30] are slightly lower than those found in this study due to the different climate conditions of the countries and the barn management.
Based on the results of the one-way ANOVA, NH 3 emissions in W1 were significantly different (p < 0.001) from those in W2 with higher NH 3 emissions in W2. As the sprinkler system in W2 was not operated, the emissions in W2 were higher due to the lack of the dilution effect of water on NH 3 . In fact, the presence of water on the floor reduces the pH of the pure urine necessary for the NH 3 formation and, consequently, the emissions decrease [31]. Moreover, even though there was not a significant difference between air temperatures in W1 and those in W2, the mean value of air temperature in W2 was slightly higher than that in W1. This is one of the main influencing factors of the increase in NH 3 emissions as it was found in previous studies [13,30].
The mean CH 4 emission rate was 5.85 g LU −1 h −1 in W1 and 5.52 g LU −1 h −1 in W2. The hourly-averaged CH 4 emission values were 2.40 to 8.83 g LU −1 h −1 in W1 and 2.89 to 9.53 g LU −1 h −1 in W2. The results of the one-way ANOVA showed that there is a significant difference (p < 0.009) between CH 4 emissions in W1 and CH 4 emissions in W2 with higher CH 4 emissions in W1 than those in W2. This result can be explained by the different management of the cooling system between the two weeks. In fact, when the sprinklers were not operated in W2, the CFI reduced from 0.28 in W1 to 0.25 in W2 and, therefore, the time spent at feeding was lower in W2 compared to that in W1 (Table 2). Conversely, the time spent lying increased in W2 with a CLI increase from 0.49 in W1 to 0.53 in W2. The different animal behaviour was related to the response to heat stress [32] that increases from W1 to W2. In detail, the THI in W2 was significantly different than that in W1 (p = 0.007) with a higher THI in W2 (Table 2). In the literature, it was found that the time spent at feeding reduces the dry matter intake (DMI) [33] and that there is a high correlation between CH 4 production and DMI [34]. Therefore, the reduction in CH 4 emissions in W2 found in this study could be attributed to a reduction in DMI in response to heat stress. Moreover, it can be highlighted that while cows generally spend more time standing than lying under stress conditions to favour heat dissipation [17,18], in the barn under study, the presence of the fogging system in the resting area increased cow comfort and, thus, the time spent lying increased [8,9]. The NH 3 and CH 4 emissions changed hourly during the day where several peaks occurred, as it was found by Ngwabie et al. [35]. Daily variations in NH 3 and CH 4 emissions in both W1 and W2 are shown in Figure 2. Whereas the trend of NH 3 emissions is similar in both weeks, the trend of CH 4 emissions in W2 is more variable than that in W1 due to the indirect effect of the cooling system on the animal feeding behaviour. In detail, NH 3 emissions reached the highest emissions during the early morning and late afternoon, whereas the lowest value occurred at 16.00 in both W1 and W2. The higher air velocity (Table 2) and the activation of the cooling system (Table 1)   CH 4 , emissions showed different peaks during W1 and W2 with lower emissions in the central hours of the day (i.e., 12.00 and 15.00) as for NH 3 . One of the main influencing factors on emissions, deriving from the utilized model, is the ratio between the differences in measured outdoor and indoor concentrations of CO 2 (C CO2in -C CO2out ) and gas pollutants (i.e., NH 3in -NH 3out and CH 4in -CH 4out ) [36]. It is recognized, in fact, that variations in gas concentrations can be influenced by several factors, such as diet, climatic parameters and barn management [10,22,37,38], producing effects on related emissions. Figure 3 shows the daily trend of the difference between NH 3in -NH 3out , CH 4in -CH 4out and C CO2in -C CO2out in both weeks. In detail, the ratio between (NH 3in -NH 3out ) and (CO 2in -CO 2out ) was low during the central hours of the day, from about 10.00 to 16.00, due to the lowering of the NH 3 concentration difference, while the CO 2 difference showed a constant trend. The same effect was detected for CH 4 , yet to a less extent compared to that of NH 3 . The similar trend of the daily differences for CH 4 and CO 2 due to their correlation [22] produced an almost constant CH 4 emission throughout the day, especially in W1 (Figure 2). Conversely, NH 3 emissions showed two peaks deriving from the trend of the NH 3 difference dampened by the trend of the CO 2 difference that was almost constant until 14.00 and higher until 21.00. The interquartile ranges showed that the values in W1 had a higher variability than those in W2 for concentration differences, whereas those ranges for emissions were uniform in the two weeks, especially for CH 4 .

Influence of THI on NH 3 and CH 4 Emissions
The results of the one-way ANOVA (Table 3) showed that groups of NH 3 and CH 4 emissions at different THIs were significantly different for both W1 and W2. In detail, NH 3 emissions at THI ≤ 72, 72 < THI ≤ 78 and 78 < THI ≤ 84 were all significantly different. The highest NH 3 emissions occurred when the THI was lower than 72, whereas they were lowest at 78 < THI < 84. Moreover, it was found a negative correlation between NH 3 emissions and THI (r = −0.70 with p < 0.001 in W1 and r = −0.65 with p < 0.001 in W2). These results are strongly related to the cow behaviour and can be explained by analysing the emitting source of the gas. In fact, when the THI was high during the central hours of the day (Table 2), cows were mainly in the lying behaviour in the cubicles where the cooling system was operated. Therefore, the chemical process for NH 3 production was reduced due to the presence of water on the floor [13] and the reduced mixing of urea and faeces performed by cow activity [28]. In addition, for CH 4 , the lowest emissions occurred at 78 < THI < 84. The reduction in CH 4 is not related to a direct effect of climatic factors on CH 4 emissions, but to an indirect effect of the animals and their metabolic activity. Besides the effect of heat stress on cow behaviour reported in the previous section, in the literature, Abeni and Galli [39] found that during heat stress, total rumination time decreased. The reduction in CH 4 could be related to the reduced rumination activity even though it was not proved a direct effect of rumination on CH 4 production [40].

Influence of Cow Activity on NH 3 and CH 4 Emissions
The statistical analyses at varying cow behaviour showed a significant difference between groups of gas emissions in both W1 and W2 (Table 4). NH 3 emissions were highest during barn cleaning due to the high NH 3 concentrations during cleaning measured by D'Urso et al. [28] during the same weeks. Moreover, Figure 3 showed that NH 3in -NH 3out increased during the cleaning of the barn in the early morning, whereas CO 2in -CO 2out concentrations varied less. Therefore, the ratio between pollutant and tracer gas contributed to enhance NH 3 emissions during floor cleaning; this is in line with other studies [29].
Besides the presence of water on the floor, the lowest NH 3 emissions, obtained during cow activity, were attributed to the dilution and flushing effects of the high air velocities during the central hours of the day ( Table 2). The effect of air velocity contributed to reduce CH 4 emissions during cow lying, together with the effect of the frequent cows' sleeping bouts during night hours.

Conclusions
A study on NH 3 and CH 4 emissions was carried out in an open-sided free-stall barn in a hot Mediterranean climate in order to test the effect of different managements of the cooling system. The aim of providing improved knowledge on the dependence of NH 3 and CH 4 emissions at varying barn managements was achieved in a period of two weeks, characterised by similar microclimatic conditions. Environmental and animal-related parameters were found to have a significant influence on gas emissions (p < 0.01) for both NH 3 and CH 4 . When the sprinkler system in the feeding alley was not operated, the values of NH 3 emissions increased from 4.73 g LU −1 h −1 in W1 to 5.90 g LU −1 h −1 in W2, and CH 4 emissions were reduced from 5.85 g LU −1 h −1 in W1 to 5.52 g LU −1 h −1 in W2. The effects of changes in the management of the barn can influence microclimatic conditions with the increase or reduction in heat stress for cows. The consequence is the adaptation of the animal behaviour to the different management conditions with the consequent variation in the gas emissions. Under hot climatic conditions, the reduction in gas emissions can derive not only from mitigation strategies, but also from heat stress conditions for the cows. This effect is to be analysed as it could produce negative effects on milk production. In conclusion, this study showed how the management of the cooling system should be supported by a behavioural analysis in order to verify the benefit on the animal and the environment. It is also useful to carry out analyses on daily emissions to provide additional information about how dairy cow activity is affected by the barn management. Further studies regarding the effects of barn management on animal behaviour and emissions per unit of milk produced may provide additional knowledge for the improvement of mitigation strategies aimed at emissions reduction.