The experiment was conducted in a laboratory (Figure 1) which consists of twelve independent bench-scale rooms (1.2 m wide × 2.4 m long × 2.4 m high), arranged side by side. Each room is equipped with a variable speed exhaust fan. The incoming air, drawn from outside the laboratory, is the same for all the rooms and comes from a main duct where it is pre-conditioned. An air conditioning unit is used, if necessary, to cool the air before entering the rooms. During the cold season, the air is heated by a heating system located just after the air conditioner. A second heating unit, located in the ventilation duct of each room, allows an optimal temperature adjustment for each chamber.

Three hundred and sixty hens (30 per room) were housed in three different housing systems (Figure 2):

Deep liquid manure pit system—Hens were confined to a commercial cage system (Ranch Cunicole G.L.R. Inc., Saint-Hyacinthe, QC, Canada) measuring 1,524 mm long, 457 mm deep and 457 mm high. The two levels, assembled as an “A” frame, included five sections of 305 mm in length with three hens in each one (465 cm² hen⁻¹). Manure dropped beneath the cages into a pit at the bottom of the room where water was added to facilitate pumping the manure in a liquid form at the end of the trials;

Manure belt system with ND—Hens were reared in Farmer Automatic’s Multi-Deck battery cages (485.5 mm wide × 507 mm deep × 540 mm high) placed 2 × 2 on three decks for a total of six cages. Each cage included five hens (492 cm² hen⁻¹). Manure dropped on a belt beneath each row of cages and was removed twice a week;

Manure belt system with FAD—Manure dropped on a belt beneath each row of cages where it was dried with forced air and removed twice a week. The drying system was installed under all the decks of the Farmer Automatic’s battery cages. A perforated 7.5-cm duct blew air from a 10-cm blower (VTX-400, Atmosphere, Terrebonne, QC, Canada) located beneath the air inlet. Five-mm holes were placed at each 160 mm with a 45-degrees angle. Then, based on an air flow of 1.3 m³ h⁻¹, the blower was adjusted to obtain an air velocity of 3.05 m s⁻¹.

Laying hens were Lohmann LSL-Lite. The lighting period was 13 h d⁻¹ for weeks 1 and 2, 13.5 h d⁻¹ for weeks 3 and 4 and 14 h d⁻¹ for the last four weeks. The lighting system was regulated to give 40 lux per room (Light Meter, Lux/FC, 840020C, Sper Scientific, Ltd., Scottsdale, AZ, USA). Water was provided by a solenoid activated valve connected to a data logger to register the water flow through the nipple drinkers inside the cages. The hens were fed 100 g hen⁻¹ d⁻¹ of a commercial diet (2005-000, Aliments Breton, St-Bernard-de-Beauce, QC, Canada).

The temperature and relative humidity of the air in each individual room were measured using a probe (model CS500, accuracy ±0.5 °C at 20 °C and ±3% RH, Campbell Scientific Canada, Corp., Edmonton, AB, Canada ). A datalogger was connected to a computer to upload data coming from the temperature-relative humidity probe every 10 s and the average value was recorded every 15 min. In accordance with commercial practices, the temperature in the rooms was set at 22.5 °C.

Ventilation rates were calculated from a 204-mm iris orifice damper (Model 200, accuracy ±5%, Continental Fan Manufacturing Inc., Buffalo, NY, USA) installed in the exhaust duct of each room. The difference of pressure was measured across the damper every 10 s and a data logger recorded the average every 15 min during each trial. The average value was then used to calculate the ventilation rate using Equation (1) as follows:

Q = k ∆ p r e s s u r e × 0.0283 1 1

where Q is the ventilation flow rate for one room during one sampling event (m³ min⁻¹), k is a constant which depends on the setting of the iris damper, Δ pressure is the difference in static pressure (inches of water) and 0.0283 is a conversion factor to obtain the desired SI units.

Manure samples were collected every week of each trial. A fixed amount of manure was taken from a random spot in each room. In the deep-pit rooms, the samples were taken from the pit under the floor. In the manure belt houses, the samples were taken on the belt. The samples were analysed for dry matter content (DMC), pH, total nitrogen (TN), ammonium nitrogen (NH₄-N) and minerals (P, K, Ca and Mg).

The sampling air was pumped to a mobile laboratory through Teflon™ tubing. In this laboratory, CO2, CH4 and N2O were analyzed by gas chromatography (GC) (model 3600, Varian, Walnut Creek, CA, USA). The strategy for chromatographic analysis was the separation of the three gases in columns packed with Porapak Q (Waters Corporation, Milford, MA, USA). The CH4 was quantified with a flame ionisation detector while the CO2 and N2O were measured with an electron capture detector. The instrumental errors on CO2, CH4 and N2O concentrations were ±30, 0.5 and 0.1 ppmv.

Samples were pumped from the experimental rooms through the injection loop of the GC for 15 min, after which the sample was analysed by the GC. A multiport valve was used to cycle gas sampling between the different points every 15 min. Concentration measurements were taken continuously during the entire experiment and were synchronized with the ventilation flow rate. Every three days between 12:00 PM and 1:00 PM, a standard gas containing known concentrations of the three GHG was analyzed to verify the response of the GC. Dust filters were installed before each trial at the inlet of each Teflon™ sampling line to avoid contamination and damage to the gas analyzers.

The GHG emissions were calculated for each sampling period using Equation (2) as follows:

E G H G = ( C o u t - C i n ) × Q N h e n s × P a t m - P v 287 × T × M G H G M a i r × 525.6 2

where E_GHG represents CO₂, CH₄ or N₂O emissions for one room during one sampling event (g yr⁻¹ hen⁻¹), C_in is the GHG concentration at the room inlet (ppmv), C_out is the GHG concentration at the exhaust fan of the room (ppmv), N_hens is the number of hens in the room, P_atm and P_v are respectively the atmospheric pressure at sea level and the vapor pressure in the room (Pa), T corresponds to the temperature (K), M_GHG characterizes the molar masses of CO₂ (44.0 g mol⁻¹), CH₄ (16.0 g mol⁻¹) or N₂O (44.0 g mol⁻¹), M_air signifies the molar mass of air (28.97 g mol⁻¹), 287 is the thermodynamic constant of air (J kg⁻¹ K⁻¹) and 525.6 is a conversion factor (mg min⁻¹ to g yr⁻¹).

The experiment was carried out over two successive 8-wk trials (March–May 2010 and June–August 2010) corresponding to the beginning of the egg laying period (19 to 27 weeks of age). This 8-wk period was found to produce the highest rate of gas emissions [25]. An 11-day acclimatization period preceded each trial. The four repetitions of each housing system were randomly assigned to the experimental rooms.

Consequently, the experiment was a completely randomized design with three housing systems and four repetitions, repeated twice. Weekly averages were calculated from all emissions measured for each period of 15 min within one week and reported in kg or g yr⁻¹ hen⁻¹ to facilitate comparison with the literature. The measurements made during the eight weeks of a same room are considered as being correlated. The analysis of variance with repeated measurements in time requires the use of a mixed model to analyze the effect of the systems on CO₂, CH₄ and N₂O emissions. The invariable effects of the model are the housing system, the week and the system interaction by week. Random effects are the trial, the trial interaction by system by repetition due to the variation between the rooms for each trial and the residual error. The procedure PROC MIXED of the SAS program (Version 8, 1999, SAS Institute, Inc., Cary, NC, USA) was used to adjust the model [26].

The outside temperature (12.4 °C for trial 1 vs. 22.3 °C for trial 2) between both trials was different due to the change of season (March–May for trial 1 vs. June–August for trial 2). This seasonal effect influenced the average ventilation flow rates as well since they were slightly greater during trial 2 (1.03 L s⁻¹ hen⁻¹ for trial 1 vs. 1.19 L s⁻¹ hen⁻¹ for trial 2). The minimum and maximum ventilation flow rates ranged from 0.13 to 1.98 L s⁻¹ hen⁻¹ for trial 1 and from 0.54 to 3.00 L s⁻¹ hen⁻¹ for trial 2. This corresponds to the recommended flow rates in Québec with 0.2 L s⁻¹ hen⁻¹ in winter and 3.3 L s⁻¹ hen⁻¹ in summer. Despite small flow differences between rooms within each trial, the average temperature in the rooms was considered constant indicating a proper operation of the environmental control equipment. The change of season only resulted in a small temperature difference between both trials (22.1 °C for trial 1 vs. 22.4 °C for trial 2).

The hens reared in deep-pit houses consumed more (P −1) and water (0.174 L d⁻¹) as compared to those housed in the rooms with manure belt systems (about 93 g d⁻¹ and 0.162 L d⁻¹) (Table 2). The different configuration of the cages (lighting angle, number of animals per cage, etc.) could explain this difference. In spite of the higher consumption, the hens in deep-pit rooms didn’t weigh more (P > 0.05) since the initial and final weights were similar between the three housing systems (1.37 and 1.64 kg, respectively). Because of the higher feed consumption, these hens had the lowest conversion rate (1.34 kg of food consumed per dozen of eggs produced). The efficiency of egg production was improved by 12% by using the manure belt housing systems instead of deep liquid manure pit systems.

The mass of manure produced by the deep-pit hens (140 g d⁻¹ hen⁻¹) was higher (P −1 hen⁻¹) and the FAD manure (61 g d⁻¹ hen⁻¹) due to the addition of water at the beginning and at the end of each trial (Table 3). This water addition also resulted in a lower DMC (P

could explain this result (Table 2). Besides, the difference in mass (P > 0.05) and DMC (P

During a study on 24 facilities in Québec, Seydoux et al. [27] noted a 12-percent DMC for liquid manures and a 55-percent DMC for ND and FAD manures. However, the reported values differ from these values. In the case of liquid manures, Seydoux et al. [27]’s samples were taken directly from the liquid manure tanks where dilution by rain could occur. This may have resulted in a lower DMC of manure samples. Nevertheless, it should be noted that a DMC of 10%, near what Seydoux et al. [27] proposed, was obtained at the beginning of each trial with the addition of water. However, it increased quickly to about 30% during the subsequent weeks possibly because of suspected leaks in the storage pit. The inability to retain all the liquid thus resulted in an average DMC of 23%, which is nonetheless comparable to what Lockyer et al. [28] found for droppings stored beneath the cages in a pit where water was added (26-percent DMC). In the case of ND and FAD manures, even though Fabbri et al. [14] obtained a DMC similar to 44% for the ventilation belt technique (28 to 42%), they considered this result lower than that generally reached with this method. Actually, the TS content of dried manure should be around 50 to 70% [9,27,28,29,30]. Higher values of DMC could have been obtained with the drying system in this study, but several short (less than 24 h) power failures occurred, probably resulting in lower DMC. Besides, some producers in Québec increase the heating during winter for greater manure drying. This operation could have influenced the results from Seydoux et al. [27].

For the deep-pit system, TN was lower while NH₄-N was clearly higher than in the other two housing systems. With manure drying, higher losses of nitrogen through NH₃ emissions should have resulted in lower concentrations in manure [25]. On the other hand, the liquid manure contained more NH₄-N because of the higher moisture content which favoured microbial conversion of urea to NH₄-N. Liquid manure analyses also resulted in lower phosphorus (P P P > 0.05) and calcium (P > 0.05) contents. Similar mineral contents were found by Seydoux’s et al. [27] for nitrogen (1.11 to 1.5 g d⁻¹ hen⁻¹), phosphorus (0.42 to 0.48 g d⁻¹ hen⁻¹), potassium (0.52 to 0.64 g d⁻¹ hen⁻¹), calcium (1.4 to 1.63 g d⁻¹ hen⁻¹) and magnesium (0.13 to 0.134 g d⁻¹ hen⁻¹).

The deep liquid manure pit and the manure belt systems with ND and FAD emitted respectively 36.0, 28.2 and 28.7 kg CO₂ yr⁻¹ hen⁻¹ (Table 4). Therefore, the ND and FAD manure belt systems reduced CO₂ emissions between 20 and 22% although 96% of CO₂ is produced by animal respiration [21]. Consequently, the significant (P 2 emissions is due to the remaining 4% attributable to manure decomposition. Long-term storage of manure inside the deep-pit rooms allows a more intensive degradation of manure into CO₂. No significant difference (P > 0.05) was found between the manure belt systems.

Furthermore, the statistical analysis showed a significant week effect on CO₂ emissions (P 2 production is directly proportional to laying hens weight, the production should have increased over time as they got heavier, as seen in Wu-Haan et al. [11]. Although the emission factors increased between weeks 3 to 7, the CO₂ production was higher during weeks 1 and 2. A possible explanation could be the change in the partitioning of energy as the birds progress from first egg to peak lay [31,32,33,34].

The mean housing CO₂ emissions were similar to the results obtained by Neser et al. [10] and Wu-Haan et al. [11] with laying hens reared in battery cages (12.61 to 37.84 and 23.95 to 32.57 kg yr⁻¹ hen⁻¹, respectively; Table 1).

The CH₄ emissions from the deep liquid manure pit, the ND and FAD manure belt systems were 31.6, 25.3 and 27.7 g yr⁻¹ hen⁻¹ (Table 4). Therefore, the manure belt systems reduced CH₄ emissions by 20 and 12%, respectively, compared to that of the deep-pit system. Since CH₄ is a product of the anaerobic digestion of manure [14,20,22], the significant difference (P et al. [35] found that an accumulation of manure generates anaerobic conditions after 25 days. In addition, there was no difference (P > 0.05) between the ND and FAD manure belt systems.

Apart from the anaerobic conditions, pH could be also a parameter responsible for a higher CH₄ production by the deep-pit system. It appears that CH₄ emissions are greatest at a pH of 7 and that acidification could be a mitigation strategy to reduce emissions [36]. In fact, the CH₄ emissions are reduced by 50% at pH 6.5. Accordingly, lower pH values (6.66 and 6.76, respectively; Table 3) for ND and FAD manures could have resulted in smaller CH₄ emission rates relative to liquid manure (7.59; Table 3). In the study of Wu-Haan et al. [11], a change in manure pH from hens fed two different diets resulted in a 17% reduction in CH₄ emissions. The slight difference in pH between the ND and FAD systems could also explain that the ND system resulted in smaller CH₄ emissions even though FAD manure had a greater DMC, which is supposed to lower emissions [14].

A significant difference in CH­₄ emissions was also found among the weeks for all the systems (P et al. [11] who found that 21-wk-old hens likely produce greater CH₄ emissions than at 38 or 59 weeks of age. The results in this study could not confirm this tendency since emissions during weeks 7 and 8 were greater than those of weeks 2 to 6 (Figure 4). In addition, the difference between the first two weeks could be, as for CO₂ emissions, the result of a change in the partitioning of energy for egg production or weight gain [31,32,33,34].

The measured values are in accordance with those of Fabbri et al. [14], Wathes et al. [16] and Wu-Haan et al. [11] who obtained CH₄ emission rates between 28.94 and 32.56 g yr⁻¹ hen⁻¹ for similar deep-pit and FAD manure belt housing systems, respectively (Table 1). However, CH₄ emissions twice that of those in this study (56 g yr⁻¹ hen⁻¹; Table 1) were observed by Groot Koerkamp and Uenk [12], Hörnig et al. [15] and Monteny et al. [13]. In the case of Hörnig et al. [15], their experiment took place in farm-scale barns, which resulted in large variations within the area. High ventilation rates could have also caused greater emissions, especially in summer where emission rates sometimes reached 320 g CH₄ yr⁻¹ hen⁻¹.

The N₂O emissions measured for the deep liquid manure pit, the ND and FAD manure belt housing systems were 2.78, 2.60 and 2.48 g yr⁻¹ hen⁻¹, respectively (Table 4). These results were statistically similar (P > 0.05). The measured N₂O emissions are below those obtained by Chadwick et al. [17], Sneath et al. [18] and Wathes et al. [16] (13 to 30 g yr⁻¹ hen⁻¹; Table 1), but comparable to Neser et al. [10] (0.63 to 4.73 g yr⁻¹ hen⁻¹; Table 1). The results obtained here come from the very low N₂O concentrations measured, only slightly higher than the concentrations in ambient air. The same observations established by some authors [14,16,37] indicated that N₂O emissions from layer houses should be judged critically since they vary considerably due to the low concentrations from which emissions are calculated. In certain cases, no significant emissions were registered for N₂O since they were consistently near the detection limit for deep-pit systems and manure belt systems [14]. Nevertheless, the very lowest N₂O emissions should be taken into account due to their potential of global warming (310 times higher than CO₂).

Total GHG emissions can be calculated on a CO₂-equivalent basis (Table 5). The evaluation of GHG emissions is based only on manure emissions. In this case, since the majority of CO₂ is produced by animal metabolism, only CH₄ and N₂O were considered. The FAD (1.35 kg CO₂-eq. yr⁻¹ hen⁻¹) and ND (1.34 kg CO₂-eq. yr⁻¹ hen⁻¹) manure belt housing systems reduced GHG emissions by 11.5 and 12.5% compared to the deep liquid manure pit houses (1.53 g CO₂-eq. yr⁻¹ hen⁻¹). These results only consider the emissions at the barn level and could vary if other factors such as energy consumption were also taken into consideration.

Name