Effects of rumen undegradable protein sources on nitrous oxide, methane and ammonia emission from the manure of feedlot-finished cattle

The effects of sources of rumen undegradable protein (RUP) in diets on methane (CH4), nitrous oxide (N2O) and ammonia (NH3) emissions from the manure of feedlot-finished cattle were evaluated. We hypothesized that the use of different RUP sources in diets would reduce N loss via urine and contribute to reduced N2O, CH4 and NH3 emissions to the environment. Nellore cattle received different diets (18 animals/treatment), including soybean meal (SM, RDP source), by-pass soybean meal (BSM, RUP source) and corn gluten meal (CGM, RUP source). The protein source did not affect the N and C concentration in urine, C concentration in feces, and N balance (P > 0.05). The RUP sources resulted in a higher N2O emission than the RDP source (P = 0.030), while BSM resulted in a higher N2O emission than CGM (P = 0.038) (SM = 633, BSM = 2521, and CGM = 1153 g ha−2 N–N2O); however, there were no differences in CH4 and NH3 emission (P > 0.05). In conclusion, the use of RUP in diets did not affect N excretion of beef cattle or CH4 and NH3 emission from manure, but increased N2O emission from the manure.

. Fecal and urinary N content and C and N balance of Nellore cattle fed with sources of rumen undegradable protein during the finishing phase in feedlots. 1 SM = manure of animals fed soybean meal as a source of rumen degradable protein (RDP), BSM = manure of animals fed by-pass soybean meal as a source of rumen undegradable protein (RUP), CGM = manure of animals fed corn gluten meal as a source of RUP; 2 SEM = standard error of the mean. Animal considered as an experimental unit (n = 9).  (Fig. 2). Highest peak of N 2 O emission was observed in the 21st day, on all treatments. On the same day, an increase in CH 4 fluxes was also observed. Differences in N 2 O and CH 4 fluxes among treatments occurred in some days of evaluation and were not consistent along the studied period.
Protein sources did not affect cumulative CH 4 emission from animal manure (P > 0.05) ( Table 2). However, the manure of animals fed RUP sources resulted in a higher cumulative N 2 O emission than that of animals fed the RDP source (P = 0.030). Emissions from manure of cattle fed CGM were almost double and emissions from manure from cattle fed BSM (P = 0.038) were quadrupled compared to SM-fed cattle.
An interaction between sampling time and protein source was observed for DM, OM, N, C and NH 4 + ( Table 3, Fig. 3). The manure of animals fed CGM presented a lower N content and higher NH 4 + than that of animals fed SM on day 42 (P < 0.001), while on day 63 higher values of N and NH 4 + were observed for the manure of animals fed CGM in relation to BSM (P = 0.002 and P = 0.010 respectively) and SM (P = 0.004 and P < 0.001, respectively). The manure of animals fed SM showed a higher C content than that of animals fed source of RUP on day 42 (P = 0.001). The manure of animals fed SM showed a higher C/N than that of animals fed RUP (P = 0.001). Nitrate content of the analyzed samples was not detectable.
A positive correlation was observed between CH 4 and C/N ratio on day 42 (P = 0.025), and between CH 4 and NH 4 + on day 63 (P = 0.001). On day 105, N 2 O was positively correlated with DM (P = 0.018) and NH 4 + (P = 0.008) ( and highest (37.3 °C) temperatures were recorded on the first sampling day and on day 19, respectively. Cumulative precipitation throughout the experimental period was 320.5 mm, occurring on 30 different days (Fig. 4). Manure from all treatments showed the highest daily mean NH 3 emission on the first day of evaluation (Fig. 5). Subsequently, NH 3 emission decreased until the fourth day of evaluation under all treatments. From the 19th day, a new peak of NH 3 emission was observed under all treatments. The SM treatment presented a small increase in NH 3 emission on days 38 and 51, while the BSM and CGM treatments presented a decrease in emission. Ammonia emission under all treatments completely ceased on the 77th day. From day 13 to 25, cumulative NH 3 emission under the SM treatment was higher than that under the BSM and CGM treatments. However, after this period, no differences were observed among the treatments.
There were no significant differences in cumulative NH 3 emission from the manure during the evaluated period and manure content of DM, OM, N, and C (P > 0.05) among the three protein sources (Table 6). Likewise, there were no differences in the C/N ratio of the manure between the RDP and RUP sources (P = 0.491). However, the manure of animals fed BSM showed a higher C/N ratio than that of animals fed CGM (P < 0.001). The manure  www.nature.com/scientificreports/ of animals fed RDP showed a higher NH 4 + concentration than that of animals fed RUP (P < 0.001); however, there were no differences in NH 4 + concentration between the manure of animals fed CGM and BSM (P = 0.670).

Discussion
Gas emissions. The use of RUP sources in the diet did not reduce N loss via urine. Meanwhile, it increased N 2 O emission but did not affect CH 4 emission from manure. Therefore, our hypothesis that RUP inclusion in the diet would reduce N loss and contribute to reduced N 2 O and CH 4 emissions from the manure was rejected. The manure deposited in the soil enhances its N and C content, thereby altering the N mineralization rate and stimulating N 2 O production 29,30 . Meanwhile, labile C released during material decomposition regulates the seasonality of N 2 O and N 2 production 30 . Inorganic forms of N (NH 4 + and NO 3 − ) are determinants of N 2 O production. The manure of animals fed BSM and CGM presented a higher NH 4 + concentration than that of animals fed SM in the 42nd day, and in 63rd only for the CGM treatment, evidencing a temporal variation of the manure characteristics in relation to treatments, probably resulting in higher urea hydrolysis at these moments ( Table 3, Fig. 3). The higher NH 4 + concentration under the RUP treatments may have promoted nitrification or denitrification, resulting in N 2 O production 17 (Table 2).
A reduction in urinary N was expected with the inclusion of RUP in diets, since an increase in N use efficiency is observed when lower RDP amounts are used in the diet 31 . In other words, reduced NH 3 concentration in the rumen was compensated via increased urea recycling to maintain an optimum ruminal NH 3 concentration for maximum microbial growth, thus decreasing its N urinary excretion 27,34,35 . However, this did not occur because the content of RDP in diets with RUP likely met the microbial demand for N, thereby not achieving sufficient urea recycling and allowing urinary N excretion similar to the diet with RDP (Table 1).
Elevated amounts of amino acids reaching the small intestine is another factor contributing to a greater N loss. When absorbed in excess or in imbalance relative to the animal's requirements, these amino acids can be oxidized for energy production, leading to urea production in the liver, which is then excreted via urine. This might occur when the diet offers adequate levels of NH 3 to meet the ruminal demand 34 . Therefore, excess CP concentration in the diet, either as RDP or RUP, may contribute to urinary N excretion.
The greatest N 2 O emission from the manure of animals fed the RUP sources (Table 3). This indicates that these diets probably had a higher urea content of the manure, since N 2 O emission is particularly affected by the urinary urea content 35 .
When higher RUP levels are used in the diet, a change in the route of urine-to-feces excretion is expected due to a higher amount of intact protein that reaches the intestine, which contributes to fecal N excretion when Table 2. Cumulative CH 4 and N 2 O emissions from the manure of Nellore cattle fed with sources of rumen undegradable protein during the finishing phase in feedlots. 1 SM = manure of animals fed soybean meal as a source of rumen degradable protein (RDP), BSM = manure of animals fed by-pass soybean meal as a source of rumen undegradable protein (RUP), CGM = manure of animals fed corn gluten meal as a source of RUP; 2 SEM = standard error of the mean. Chamber considered as an experimental unit (n = 9). The cumulative values refer to 112 days of feedlot.  Table 3. Chemical composition of the manure, deposited in the soil, of Nellore cattle fed with sources of rumen undegradable protein during the finishing phase in feedlot. 1 DM = dry matter (g kg −1 ), OM = organic matter (g kg −1 ), N = nitrogen (g kg −1 ), C = carbon (g kg −1 ), NH 4 + ammonium (mg kg −1 ); 2 SM = manure of animals fed soybean meal as a source of rumen degradable protein (RDP), BSM = manure of animals fed by-pass soybean meal as a source of rumen undegradable protein (RUP), CGM = manure of animals fed corn gluten meal as a source of RUP.; 3 SEM = standard error of the mean; 4 T = time; 5 Interaction TR (treatments = SM, BSM and CGM) × T (time). Chamber considered as an experimental unit (n = 9). www.nature.com/scientificreports/ not absorbed 38,39 . However, there were no differences in fecal N excretion between the RDP and RUP treatments (Table 1), although fecal N concentration differed between the two RUP sources. This might be attributed to the distinct amino acid composition or the different chemical structures of these sources. The processes through which corn (corn gluten, a by-product of wet corn milling) and soybean (thermally treated) have been subjected can make the protein undegradable in the rumen or unavailable 38 . Despite different compositions of the manure among the treatments (Table 3), there were no differences in CH 4 emission ( Table 2). Nitrogen and OM contents and C/N ratio of manure are important factors associated  www.nature.com/scientificreports/ with CH 4 emission 41,42 . Nevertheless, differences in manure chemical composition among the treatments were observed in some sampling days (Fig. 3). This result can be related to variations in environmental conditions, such as temperature and precipitation, which can alter the chemical composition of manure. However, these differences among the treatments were not consistent throughout the experimental period, justifying the lack of differences in CH 4 emission.
In manure, most of the N content comes from N excreted via urine in the form of urea, which is rapidly hydrolyzed to NH 4 + , and N losses from organic forms of feces also occur 41 . Organic N can promote CH 4 emission, playing an important role in the transformation of acetate to CH 4 42 , whereas mineral N as NH 4 + can inhibit CH 4 production, breaking the link between acidification and methanogenesis in anaerobic processes 43 .
Nitrous oxide and CH 4 fluxes varied from −62 to 318 µg N 2 O m 2 h −1 and from −125 to 321 µg CH 4 m 2 h −1 , respectively, during the experimental period (Fig. 2). These fluxes showed a great variation, which can be attributed to several factors, such as the temporal variation in the chemical composition of manure due to variations in climatic conditions, as explained above (Table 3, Fig. 3). Other researchers 44 have reported a large variation in emissions, mainly associated with irregular fecal and urine deposition on the surface, which may also have occurred in the present study.
Frequent deposition and accumulation of feces and urine in the soil did not increase CH 4 and N 2 O emissions over time under all treatments. Trampling by animals may have caused aeration of the surface material and have provided unfavorable environment for the action of methanogenic bacteria and nitrifying/denitrifying microorganisms. In addition, the humidity in the feedlot did not increase over time, based on the DM content of the manure, except on rainy days (Table 3). This is probably related to the dry climate at that time of year, associated with the compacted soil of the feedlot. Table 5. Pearson's correlation coefficients between explanatory variables on each sampling day. *Represents a statistical significance (P ≤ 0.05) for the coefficients of correlation. 1 DM = dry matter, OM = organic matter, NH 4 + = ammonium. www.nature.com/scientificreports/ Precipitation and temperature changes strongly affect CH 4 emission 45 . During the study period, CH 4 flux was related to these variables. On the 21th day, increased emission peaks were observed under all treatments, probably due to precipitation in the previous week. Considering that CH 4 emission occurs under anaerobic conditions, precipitation may have favored higher emissions due to increased moisture content of the manure 46 . On the 49th day, reduced CH 4 emission was observed, possibly due to temperature drop on that day. Considering that CH 4 emission is a biological and anaerobic process, temperature can act as a limiting factor by reducing methanogen activity 47 . After this period, CH 4 emission tended to stabilize, probably due to the absence of high precipitation and little variation in temperature (Fig. 1).
The mean CH 4     www.nature.com/scientificreports/ study was lower than reported values by other researchers (mean of 110 g day −1 pen −1 ) 48 under similar climatic conditions and a pen density of 6 m 2 per animal, however, the floor was concreted and the excreta were removed every 15 days. The low moisture of the manure was possibly responsible for low CH 4 emissions, because even under favorable chemical conditions, microbial activity is limited at low moisture levels. Of note, the density in each pen was 30 m 2 per animal and the evaluations were performed near the feeders, in an area of 6.5 m × 10 m with higher excreta deposition. The density of animals is reflected in the condition of excreta deposition and accumulation on the surface, and it is a relevant factor to be considered when evaluating gas emissions in feedlots 49 . On some sampling days, CH 4 uptake occurred predominantly through the consumption of atmospheric CH 4 , which can occur in aerobic environments 39 . The environment is a CH 4 source when the balance between methanogenic production and methanotrophic consumption is positive, leading to CH 4 emission. In contrast, when this balance is negative, the environment is considered a CH 4 sink 39 .
Considering that the feedlot system has emerged as a management strategy to minimize the impacts of lower forage production in the dry season, majority of feedlots in Brazil are managed from April to November, when rainfall is scarce and temperature is low. The climatic conditions during this period, when associated with feedlots of low animal density, can result in low CH 4 emission. In an inventory to estimate GHG emission in Brazil 50 , it is clear that we do not have enough data to estimate emissions from Brazilian feedlots. Therefore, measurements must cover different systems, with different stockings, feedings and manure management to generate concrete data that allow the comparison between mitigation strategies.
The Post-rainfall emissions and wetting of the area might be related to a combination of mineralization, nitrification, and/or denitrification, leading to the release of N 2 O absorbed in the dry soil 52 . Moisture is an important factor in N 2 O production, particularly when associated with temperature and a propitious chemical composition 53 , emission of N 2 O increases markedly with increasing temperature 54 . However, after reaching the peak, N 2 O emissions remained stable, with small variations across evaluation days; even in the presence of additional precipitation events, low temperature (minimum of 3.3 °C near the 49st day) may have hampered the occurrence of new emission peaks.
Nitrate was not detectable in the manure during the experiment. Nitrous oxide production is assumed to occur through nitrification, via the oxidation of NH 4 + in hydroxylamine (NH 2 OH), with NOH as an intermediate and N 2 O as the product 55 . N 2 O can also be produced through denitrification by nitrifiers, wherein NH 3 is nitrified and oxidized to nitrite (NO 2 − ), which is then reduced to nitric oxide (NO), N 2 O, and molecular N (N 2 ). Nitrous oxide is an intermediate in the reduction of NO 2 − to N 2 56 . During denitrification, NO 3 − is used as the primary substrate 57 . Denitrification may not have occurred in the present study.
Correlation analyses showed no significant linear associations of CH 4 and N 2 O production with the tested variables related to the chemical composition of manure, which can be attributed to specific factors (Tables 4  and 5). The processes underlying the production of gases are complex and rely on the chemical composition of manure. In addition to the chemical composition, the emission of gases in the manure is dependent on other factors such as temperature, moisture, deposition conditions, and trampling by animals. The absence of significant correlations between gas production and manure composition may be related to the small variation in the characteristics analyzed during the sampling period, making it difficult to observe relationships among variables. NH 3 emission. The use of RUP in the diet did not reduce N loss via urine and did not influence NH 3 emission from the manure. In this sense, our hypothesis that RUP inclusion in the diet would reduce N loss and contribute to decreased NH 3 emission was rejected.
The manure of animals fed SM presented higher NH 3 emissions than that of animals fed CGM and BSM from the 8th to 25th day of evaluation. This may be attributed to the higher NH 4 + content of the manure of animals fed SM than that of animals fed CGM and BSM at the beginning of the sampling period (Table 6). Subsequently, the manure of animals fed CGM and BSM presented a new NH 3 emission peak following the event of the highest precipitation (54.2 mm) throughout the experimental period. However, during this period, most of the NH 4 + from the SM treatment had already been used, as reflected by the weak response to precipitation under this treatment. Urea present in the urine and feces is rapidly hydrolyzed, and the formed NH 4 + is dissociated to aqueous NH 3 , depending on NH 4 + concentration and pH of manure and environmental conditions. When precipitation occurs, urease activity is promoted, resulting in increased NH 3 emission 58 . Of note, however, manure sampling for characterization was performed before implanting the chambers in the area. Thus, the chemical composition data presented herein do not represent the possible temporal variations during the NH 3 evaluation period (Table 3).
Higher values of NH 3 emission have been reported (49.1 kg NH 3 animal −1 ) in beef cattle feedlots, which is mainly related to the fact that the majority of confinement feedlots are outdoors, given that wind speed in open environments increases emission 59 . According to others studies 19 , daily NH 3 emission in feedlots rarely exceeds 2000 µg NH 3 m −3 ; however, in the present study, higher values were observed. Importantly, as explained before, the evaluations were performed in an area of higher excreta deposition, with the objective of comparing the treatments in homogeneous conditions of excreta distribution. Therefore, the amount of emission by area of the total feedlot may have been overestimated in this study. Conversely, we did not account for emissions when the animals were present in the feedlots. Throughout the sampling period, the animals had already been removed from the area, and there was a large amount of accumulated manure. When the wet season starts, emission may www.nature.com/scientificreports/ have been favoured by increased moisture content due to the large amount of available substrate 19 . Therefore, the urea excreted by the animals was hydrolyzed and contributed to the stock of NH 4 + , which was emitted as NH 3 when the moisture content increased as a function of precipitation.
Over time, as no new manure was deposited due to the absence of animals in the area, emission probably ceased when the substrate was consumed, which occurred around the 77th day in the present study. In experiments in which excreta from the animals is collected and then applied to the soil for evaluation in the absence of animals and new depositions, ammonia emission occurs for 3 weeks on average 42,62,63 . Therefore, further studies are warranted to investigate NH 3 emission in open feedlots and to observe peaks occurrence in the presence of animals, maintaining the evaluations also after removing the animals, in the next rainy season.

Conclusions
The inclusion of RUP in the diet did not affect N excretion by animals. While the N 2 O emission from the manure was increased, CH 4 emission and NH 3 emission remained unaffected. Additional studies are warranted to investigate the effects of using different proportions of RDP and RUP in diets on NH 3 , N 2 O, and CH 4 emissions from the manure of animals managed in feedlot systems under tropical conditions.

Material and methods
The experiment was approved by the Ethics, Bioethics, and Animal Welfare Committee of São Paulo State University (UNESP), Jaboticabal, under protocol numbered 16.668/16. All methods were carried out in accordance with relevant guidelines and regulations. Methods are reported in the manuscript following the recommendations in the ARRIVE guidelines.
Site description. The present study was conducted at the Campus of Jaboticabal of the São Paulo State University, Sao Paulo, Brazil (21°14′05″S, 48°17′09″W; altitude, 615.01 m). The region has a tropical climate, with a dry season from April to September and a wet season from October to March, during which over 80% of the annual precipitation occurs. The soil is Rhodic Ferralsol 62 derived from basalt, with a sandy-clay-loam texture (10% silt and 61% sand) in the surface layer (0-10 cm). The soil pH in CaCl 2 is 5.9, bulk density is 1.8 kg dm −3 , and organic matter content is 16.6 g dm −3 at the same depth.
Meteorological data (daily precipitation and ambient temperature) were obtained from the dataset of the Agrometeorological Station of the Department of Exact Sciences, Universidade Estadual Paulista (UNESP), Campus of Jaboticabal, located 1 km from the experimental area.
Experimental design. The experiment was conducted for 210 days from May to December 2019. The first 21 days were dedicated to animal adaptation to the diet, followed by 112 days of confinement, during which weekly sampling of N 2 O and CH 4 was performed. After removing the animals from the feedlots, NH 3 was sampled for 77 days.
Fifty-four Nellore bulls with an initial body weight of approximately 360 kg were distributed in three treatments. The animals were divided into three treatments and allocated in collective pens (11 m × 50 m; one pen per treatment and 18 animals per pen). Each pen had a dirt floor with collective drinkers for every two pens. There were two covered automated feeders in each pen (INTERGADO®, Intergado Ltd., Contagem, Minas Gerais, Brazil). The feed system was equipped with an automated feeder monitor resting on load cells, allowing electronic registration of the amount of feed consumed by animal. The trough recognizes the animal from the electronic ear tag, automatically sends consumption data to a database, and stores the information.
Manure of animals fed with sources of protein (two sources of RUP and one source of RDP as a control) was collected, resulting in three treatments as follows: (1) Soybean meal (SM): source of RDP. The experimental diets were composed of 30% roughage and 70% concentrate, formulated to meet the average daily gain (ADG) of 1.5 kg day −1 , according to BR CORTE 63 . The diets were offered at 08:00 am and 04:00 pm. The amounts offered were sufficient to allow a daily leftover of 5-10% of the total offered.
The ingredients of the diets were analyz ed for chemical composition ( Table 7). The AOAC 64 method was used to determine dry matter (DM) (method 930.15), crude protein (CP) (method 990.03), organic matter (OM) (method 942.05), and ether extract (EE) (method 920.39) content. Neutral detergent fiber (NDF) content was determined according to the method described by 65 using ANKOM® 2000 (Ankom Technologies, New York, USA) with thermostable α-amylase and without sodium sulfite, corrected for ashes and residual proteins. The RDP and RUP content was estimated based on the protein fraction 66 and degradation rate of each fraction, considering a passage rate of 5% h −1 .
Gases (N 2 O, CH 4 and NH 3 ) were sampled using chambers (n = 9 per treatment) arranged in an area of 65 m 2 , near the feeders, where the manure (feces and urine) was deposited the most frequently. The chambers were placed on manure (feces and urine) that had been deposited on the feedlot surface by animals subjected to treatments. At the time of evaluation, the chambers were randomly placed in an area (6.5 m × 10 m) delimited near the feeders inside each confinement pen. Specifically, an area of higher excreta deposition was selected with the objective of treatment comparison, thus avoiding evaluation in places without homogenous excreta distribution (Fig. 6).  www.nature.com/scientificreports/ out between 4:00 pm and 04:00 pm. The chambers were closed for 20 min, and air samples were collected at 0, 10, and 20 min using a 50 mL polypropylene syringe and then transferred to previously evacuated chromatography flasks (20 mL). The temperature inside and outside the chamber was measured using a digital thermometer (Incoterm®) to correct gas fluxes. Air samples were analyzed using gas chromatography (Shimadzu Greenhouse Gas Analyzer GC-2014; Kyoto, Japan) under the following conditions: (1) N 2 O: injector temperature, 250 °C; column temperature, 80 °C; N 2 carrier gas (30 mL min −1 ); and electron capture detector temperature, 325 °C; and (2) CH 4 : H 2 flame gas (30 mL min −1 ) and flame ingestion detector temperature, 280 °C.
Nitrous oxide (µg N-N 2 O m −2 h −1 ) and CH 4 (µg C-CH 4 m −2 h −1 ) fluxes were calculated considering a linear increase in gas concentration inside the chamber during the closed period and corrected for ambient temperature, ambient pressure, and chamber dimensions, as follows: where gas is the increment in the gas concentration inside the chamber during the closed period (ppb min −1 ); Mω is the molar mass of N-N 2 O (28 g mol −1 ) or C-CH 4 (12 g mol −1 ); V is the chamber volume (m 3 ); 60 is the conversion factor from minutes to hours; 10 -6 represents the conversion factor from g to µg; A is the chamber area (m 2 ); VM corr is the molecular volume corrected by the normal conditions of temperature and pressure at the time of sampling; and 10 -9 is the conversion factor from ppb to µL m −3 .
Fluxes were multiplied by 24 to obtain daily emissions, and the daily values were integrated through linear interpolation to obtain cumulative emissions during the evaluated period. Negative fluxes were included in the calculations to avoid biased data 68 .

Evaluation of NH 3 emission.
After removing the animals from feedlots, NH 3 volatilization was evaluated until the NH 3 emission ceased by sampling on days 1, 2, 3, 4, 6,8,13,19,25,31,38,44,51,59,68 and 77 after positioning the chamber. The chambers were randomly placed above the manure (feces and urine) in the previously delimited areas. Quantification was performed according to the methodology of static chamber 69  Estimation of fecal and urinary production and N balance. Fecal production was estimated using the internal marker technique 70 based on the indigestible NDF (NDFi) marker. Fecal sampling was performed from the 60th day after the animals entered the feedlots, for three consecutive days, directly from the rectum of the animals. Sampling was performed in the morning, middle of the day, and afternoon on the first, second, and third days, respectively. A composite fecal sample, by animal (9 animals/treatment), were made with the samples from these three days. The samples were mixed, homogenized, partially dried in a forced-air ventilation oven at 55 °C for 72 h, and milled in a mill with a 2 mm sieve. Samples of the ingredients of the animals' diets were collected, and their consumption was determined using the INTERGADO®.
Fecal NDFi content was determined after incubating the samples in situ for 288 h 71 followed by extraction with neutral detergent using an autoclave 72 . Fecal DM production was determined as the ratio of the concentration of the internal indicator ingested by the animal and its concentration in feces 73 .
Urine samples were collected simultaneously with fecal samples. In brief, 50 mL aliquots of urine were sampled ("spot" sample) during three consecutive days 74 . Creatinine concentration in the spot sample was determined with a colorimetric method using a commercial kit (Labtest®). Urinary excretion was estimated based on the association between creatinine excretion and body weight using the equation proposed by 75 : where UCE = urinary creatinine excretion and BW = body weight in kg.
The fecal and urine samples were analyzed for total C and total N content using the dry combustion method with Leco® CN-828 (Leco Corporation). Nitrogen retention (NR) was expressed in grams per day and in percentage of NC, and fecal and urinary N excretion was expressed as the percentage of the total material excreted. The following equation was used to calculate NR: where NC = N consumption, EFN = excretion of fecal N, and EUN = excretion of urinary N.
Statistics. All statistical analyses were performed using SAS 9.4 (SAS Inc., Cary, NC). Response variables were analyzed in a completely randomized design using the PROC MIXED procedure. There were nine experi- www.nature.com/scientificreports/ mental units per treatment. Mean values were compared using orthogonal contrasts (SM vs. RUP and BSM vs. CGM) at a 5% probability level. Total N, total C, and C/N in feces and urine and N balance were analyzed considering a model including the treatments (SM, BSM, and CGM) as fixed effects, animals (experimental unit in the RANDOM SAS option) and residual random error (NIID) of (0, σ 2 ) as random effects.
Cumulative N 2 O, CH 4 and NH 3 emissions, and manure characteristics (DM, OM, N, C, C/N, NH 4 + , and NO 3 − of manure, sampled on day 0, before the beginning of NH 3 emissions measurements) were analyzed considering a model including the treatments (SM, BSM, and CGM) as fixed effects, chamber (experimental unit in the RANDOM SAS option) and residual random error (NIID) of (0, σ 2 ) as random effects.
Nitrous oxide and CH 4 daily fluxes and manure characteristics (DM, OM, N, C, C/N, NH 4 + , and NO 3 − , sampled on days 42, 63 and 105 of N 2 O and CH 4 evaluation) were analyzed using a repeated measures mixed model over time including the treatments (SM, BSM, and CGM), collection period and interaction as fixed effects, chamber (experimental unit and RANDOM SAS option) and residual random error (NIID) of (0, σ 2 ) as random effects. Distinct covariance matrices were evaluated and the best structure was selected according to the Akaike information criterion (AIC).
Pearson correlation analysis between gas emission (N 2 O and CH 4 ) and chemical composition (N, C, C/N, DM, OM, and NH 4 + ) of the manure was performed separately for each sampling day (days 42, 63 and 105 of manure evaluation), and also considering all data collected on these days.