Grazing Intensity Impacts on Herbage Mass, Sward Structure, Greenhouse Gas Emissions, and Animal Performance: Analysis of Brachiaria Pastureland

: A 7 year experiment (2008–2014) evaluated cattle grazing intensity (sward height) e ﬀ ects on herbage mass, forage quality, and greenhouse gas emissions in continuously stocked pastures containing the tropical ‘Marandu’ palisade grass ( Brachiaria brizantha (Hochst. ex A. Rich) Stapf cv. Marandu). The experiment consisted of three sward height treatments (15, 25, and 35 cm) and six replicates. There were four periods each year during the rearing phase. Signiﬁcant e ﬀ ects were found for herbage mass, proportions of leaf and stem, crude protein, neutral detergent ﬁber, acid detergent ﬁber, lignin, animal performance, enteric methane (CH 4 ), and greenhouse gas (GHG) emissions from soils. When the canopy height increased from 15 to 35 cm, the herbage mass rose from 5.23 to 9.86 kg t ha − 1 , leaf percentage decreased, and stem percentage increased. Crude protein content averaged 14.2%, and neutral detergent ﬁber averaged 58%. Average daily gain averaged 0.67, 0.81, and 0.90 kg − 1 head − 1 , while live weight gain ha –1 was 649, 530, and 439 kg for the 15, 25, and 35 cm treatments, respectively. The weather variables explained the GHG emissions, interannual herbage mass, and structure variations. A.d.S.C., D.R.C., R.P.B., M.V.A., A.A.O., J.F.W.K., and L.d.F.B., data collection; A.d.S.C., F.O., and A.C.R. data analysis; A.d.S.C., A.C.R., and R.A.R. writing—original preparation, review, and editing. All have read and agreed to the published version of the


Introduction
Beef cattle production is an important source of protein, minerals, and vitamins for human nutrition. However, it requires significant quantities of natural resources and may negatively impact the environment through greenhouse gas (GHG) emissions [1][2][3]. Beef cattle production is pasture-based with pasture height. The analysis used data from a 7 year experiment on a tropical grass; Marandu palisade grass.

Experimental Site and Grazing Management
The data originated from a series of experiments carried out from 1 December 2007 to 31 March 2014. The experimental site was located within the Forage and Grasslands sector at São Paulo State University, "Julio de Mesquita Filho" (UNESP) (Jaboticabal, São Paulo, Brazil). The climate in this region is classified as a humid subtropical type with wet summers and dry winters. The average annual rainfall is 1424 mm, and the average air temperature is 22.3 • C. Rainfall and air temperature were measured during the experiment ( Table 1). The soil is a Rhodic Ferralsol [16]   The pasture was fertilized on a regular basis according to the recommendations from the Committee for Fertilization and Liming of São Paulo State after considering the soil chemical analyses and forage production targets [17]. In the studied area, fertilization occurred during the rainy season in the Months of November to March. The fertilizer application schedule is summarized in Table 2 based on soil chemical analyses conducted every year.
In the studied region, the forage growing season began in October and finished in April. During the dry period, the animals remained in the field and received silage and concentrates, or the stocking rates were reduced by the farms. In the period of May to September, 11% of the slaughtered animals were finished in the feedlots. In the present study, the pasture was continuously stocked with young bulls (Bos indicus var. Nellore), and Nellore crosses with other B. indicus. The animals arrived at the experimental area in October and November. In November, the pastures achieved the desired sward height. In general, the experiments commenced in December after 15 d of adaptation and were conducted for 84 or 112 days (3 or 4 periods of 28 day).
The bulls used in the experiments were 10-12 months old and weighed 180-230 kg when the evaluations began. They were identified, weighed, and randomly distributed in groups of six bulls "testers" per paddock (108 experimental animals). Some animals (40-45 put and take animals) were used to maintain the sward height using the put-and-take method [18]. In the put-and-take method, there are fixed animals "testers" and put and take animals used to adjust the stocking rate according to the herbage accumulation (mass production) to keep the management target (sward height in this study). The height/herbage mass relationship (bulk density) was used weekly to adjust the stocking rate by use of the put and take animals.

Experiments and Treatments
From 2008 to 2014, the treatments were three sward heights (15,25, and 35 cm) in a completely randomized block with six replications per treatment. The factor of blocking was soil topography.

Response Variables
Herbage mass, sward canopy structure, herbage nutritive value, and animal performance were measured between 2008 and 2014; forage intake was quantified between 2007 and 2014; enteric CH 4 was measured between 2012 and 2014; and soil GHG emissions and other explanatory variables were quantified in 2013 and 2014. The source of the data was the nine dissertations or theses listed in the Supplementary Information SI.
Grazing heights were measured weekly using a graduated rule at 80 random points ("hits") per paddock and the data were used to estimate average sward height. The forage mass was estimated by clipping eight 0.25 m 2 quadrats at average paddock sward height locations in each pasture to a residual height of 5 cm every 28 d during the measurement period. The forage samples were separated into dead material, stem + leaf sheaths, and live leaves. Then they were dried at 55 ± 5 • C to constant weight to estimate dry matter (DM) ha −1 . The forage nutritive value, was estimated by hand-plucking method, samples every 28 d (20 samples per plot). These were then dried at 55 ± 5 • C to a constant weight and ground through a 1-mm screen in a shear mill (Thomas-Wiley Laboratory Mill Model 4, H. Thomas Co, São Paulo, Brazil.) for subsequent analyses.
Forage CP, NDF, ADF, Ether Extract, and lignin concentrations were determined using the standard procedures outlined by the AOAC (1990). Indigestible NDF was quantified by incubating hand-plucking and feces samples dried at 55 ± 5 • C to a constant weight and ground through a 2 mm for 244 in the rumen. The following equation was used to determine the apparent digestibility of DM: digestibility (%) = 100 − (100 × (iNDF in hand-plucking sample/iNDF in feces) × (nutrient concentration in feces/nutrient concentration in feedstuff) Enteric CH 4 was measured using a tracer gas method (sulfur hexafluoride-SF 6 ) in 2012, 2013, and 2014. The complete methodologies are described in Barbero et al. (2015) [19].
Animal performance variables were calculated as follows [18]: Body weight gain-BW gain (kg) = (final BW − initial BW); Average daily gain-ADG (kg animal −1 day −1 ) = (BW gain (kg)/days); Gain per area-GPH = (ADG × number of animals per days × experimental period (days))/area (ha), where the number of animal days (animal day −1 ha −1 ) was calculated by dividing animal stock by the mean weight of "testers". Animal stock was determined by the sum of weights of all animals present in each paddock divided by the area of the paddock (kg BW ha −1 ); Stocking rate in animal unit (AU = 450 kg BW)/ha = ( BWmean/450)/area (ha); Greenhouse gases from the soil were quantified using the static closed chamber method described by Cardoso et al. (2017) [12]. The explanatory variables for soil moisture, ammonium, and nitrate were quantified at the 0-20 cm soil depth. Soil samples of the 0-10 cm layer were sampled at each air collecting event to quantify mineral N, gravimetric water content (by drying soil at 105 • C), and percent water filled pore space (WFPS). Soil bulk density in the 0-10 cm layer was determined using a cylinder 50 mm in diameter and 50 mm in height. WFPS was calculated from the gravimetric water content and bulk density using a particle density of 2.65 g cm −3 . For mineral N analysis, extraction with 2 M KCl was done on field moist samples with correction for water content. Ammonium-N was determined using the Berthelot reaction with spectrometry at 647 nm. Nitrate-N quantification was carried out by ultraviolet absorption spectrometry at 220 [12]. Air temperatures, moisture, precipitation, and solar radiation were obtained from a meteorological station located 500 m away from the study area.

Statistical Analysis
Data were analyzed as repeated measures using the LME procedure in R (package NLME, R Core Team, https://svn.r-project.org/R-packages/trunk/nlme/). The model included sward height, years, and sward height versus year interactions as fixed effects, and paddock or animal as the random effect. Non-structured covariance was used for the repeated-measures analyses because it had the lowest Akaike Information Criterion. Orthogonal polynomial contrasts were performed when a significant effect was found.
Principal component analysis (PCA) was obtained from a correlation matrix using the R package "FactoMineR" (http://factominer.free.fr/). The categories analyzed were: (1). Forage mass and structure (Dry matter mass, leaf mass, stem mass, dead material mass, % of leaf, % of stem, % of dead material, and leaf/stem ratio), and (2) Nutritive value (Organic Matter%, CP%, NDF%, and lignin% of DM), In this study, canonical correlation analysis (CCA) was used to correlate monthly observations of climate independent variables (air temperature, air moisture, rainfall and solar radiation) with dependent variables: herbage mass and structure (Dry matter mass, leaf mass, stem mass, dead material mass, % of leaf, % of stem, % of dead material, and leaf/stem ratio). And, for the GHG from soil analysis, the climate variables (mean, minimal, and maximal air temperature; mean, minimal, and maximal air moisture; rainfall; and solar radiation) were the independent variables and CO 2 , N 2 O, and CH 4 fluxes were the dependent variables. Daily data were used. The best multiple regression model was chosen using R 2 and the significance of the model was calculated by the "FactoMinerR" package in R (R Core Team, 2016).

Forage Mass and Structure
Pasture height affected herbage mass (p < 0.0001), leaf mass (p < 0.0001), stem mass (p < 0.0001), dead material mass (p < 0.0001), % of leaves in the herbage (p < 0.006), and % of stems (p < 0.0158), but not the percentage of dead material (p = 0.29). Herbage mass varied from 3460 DM ha −1 in 2009 in the 15 cm treatment to 11,000 kg DM ha −1 in 2013 in the 35 cm treatment. During the seven years of evaluation, herbage mass averaged 5230, 7810, and 9860 kg DM ha −1 for the 15, 25, and 35 cm treatments, respectively. Herbage mass increased linearly with sward height.
The lowest leaf mass was recorded in 2012 (1330 kg DM ha −1 ) and the highest in 2013 (4240 kg DM ha −1 ) ( Table 3). The leaf mass range was 1760, 2460, and 3010 kg DM ha −1 for the 15, 25, and 35 cm sward heights, respectively. The average stem mass at 35 cm height (3560 kg DM ha −1 ) was more than double the stem mass recorded at 15 cm (1490 kg DM ha −1 ) and was greater than at 25 cm (2410 kg DM ha −1 ). The greatest mass of dead material was found in 2008 and the lowest in 2013. The dead material mass averaged 1940, 2830, and 3220 kg DM ha −1 , respectively, for the 15, 25, and 35 cm sward heights (Table 3). Leaf, stem, and dead material mass increased linearly with sward height.
The forage morphological composition results showed that the average leaf proportion decreased linearly from 40% to 35.7% as sward height increased (p = 0.006); the proportion due to stems increased linearly (p = 0.0158) from 42.6% to 45.2%; and the % of dead material decreased linearly, with averages of 37.2, 35.5, and 32.6% for the 15, 25, and 35 cm sward heights, respectively, ( Table 3).

Nutritive Value
The nutritive value results showed that sward height affected the organic matter (OM), NDF, and ADF values (p < 0.01). The effect of grazing intensity on CP and digestibility had p-values of 0.06 and 0.0548, respectively. Table 4 shows that OM was similar across all treatments (approximately 91%). Crude protein varied greatly among years and decreased linearly with sward height when all the year data were analyzed together. The highest CP occurred in 2011 (16.3%) in the 15 cm treatment, whereas the lowest CP content was 11.2%, which occurred in 2014. The CP content averaged 14.7, 14.0, and 14.0% for the 15, 25, and 35 cm sward heights, respectively.
Both the NDF and ADF fractions increased linearly with sward height. The highest NDF concentration was measured in 2009 for all treatments and the lowest in 2011. The average NDF values during the 7 years of evaluation were 57.8, 59.1, and 59.6% for the 15, 25, and 35 cm treatments, respectively. The ADF value for the 35 cm treatment (32.0%) was significantly higher than 15 and 25 cm sward heights, which both averaged 30.4%. Lignin content averaged 3.7, 3.8, and 4.2% for the 15, 25, and 35 cm treatments, respectively ( Table 4). The results of principal component analysis identified that temperature and rainfall explained more than 80% of the variation in Marandu palisadegrass chemical composition (temperature explained 45% and rainfall 41%). The canonical correlation analysis between the weather variables and nutritive value showed that rainfall and temperature were the most correlated variables. The coefficients were 0.23 and 0.49 for rainfall and temperature, respectively. The variation in the NDF content was explained by the variation in the temperature. The coefficient were 0.94 between NDF and temperature.

Animal Performance
Sward height affected daily weight gain (p = 0.027), which increased linearly by 0.668, 0.892, and 0.902 kg Body weight (BW) day -1 on average for the 15, 25, and 25 cm treatments, respectively ( Table 5). The highest individual gain exceeded 1 kg BW day -1 and was observed in 2013, while the lowest was recorded in 2008. The highest gain per area was recorded in 2011 and 2013 at the highest grazing intensity (15 cm sward height), and the lowest was observed in 2008 in the 35 cm treatment (196 kg BW ha −1 ). Gain per area averaged 649, 530, and 439 kg BW ha −1 and was affected by sward height (p = 0.004). This gain was recorded during the rainy season (from December to April). Stocking rate was significantly affected by sward height (p < 0.0001) and decreased linearly in response to pasture height. The stocking rate averaged 6.13, 4.73, and 4.09 animal units ha −1 (1 animal unit equals 450 kg BW) for the 15, 25, and 35 cm treatments, respectively.

Enteric Methane
Daily CH 4 emissions per animal were affected by sward height (p = 0.002) and year (p < 0.0001), but there was no interaction sward height and year. They increased linearly and per kg of BW gain (p = 0.012) e varied within year (p = 0.0003). In 2013, CH 4 emissions decreased with increasing average daily gain, whereas, an inverse effect was observed in 2014. A summary of the enteric CH 4 data is shown in Table 6. Enteric CH 4 averaged 117, 126, and 139 g animal −1 day −1 for the 15, 25, and 35 cm treatments, respectively, and the emissions per gain (g kg −1 BW gain) followed a similar pattern to the daily emissions.

Greenhouse Gases
Soil GHGs were measured from 2012 to 2014 in the long-term study and the results showed that grazing intensity affected GHG emissions. Nitrous oxide decreased, CO 2 increased, and CH 4 showed a curvilinear response to increased sward height. Only the highest grazing intensity (15 cm sward height) produced net N 2 O emissions. Furthermore, N 2 O consumption increased with sward height ( Table 7). The lowest CH 4 emissions were recorded in the 25 cm treatment, and the results showed that sward height caused a CO 2 emissions increase from 44.2 to 80 Mg CO 2 ha −1 , which was the largest sward height effect within the analyzed gases (r 2 = 0.95). The principal components analysis for GHGs emissions and explanatory variables identified that there was a clear positive correlation between soil moisture, and CO 2 and N 2 O emissions. Soil ammonium and nitrate were also correlated with these gases. However, soil CH 4 was not correlated with the variables (Figure 1). The CCA was performed to identify linear correlations between the grouping identified in the PCA analysis. The CCA showed the correlation between N 2 O and CO 2 of 0.38 and with moisture of 0.48. The correlation with ammonium and nitrate was negative −0.53 and −0.49, respectively.
Agronomy 2020, 10, x FOR PEER REVIEW 10 of 18 that sward height caused a CO2 emissions increase from 44.2 to 80 Mg CO2 ha −1 , which was the largest sward height effect within the analyzed gases (r 2 = 0.95).
The principal components analysis for GHGs emissions and explanatory variables identified that there was a clear positive correlation between soil moisture, and CO2 and N2O emissions. Soil ammonium and nitrate were also correlated with these gases. However, soil CH4 was not correlated with the variables (Figure 1). The CCA was performed to identify linear correlations between the grouping identified in the PCA analysis. The CCA showed the correlation between N2O and CO2 of 0.38 and with moisture of 0.48. The correlation with ammonium and nitrate was negative −0.53 and −0.49, respectively.

Sward Structure
This study aimed to show how sward structure affects forage quality. Despite the difference in herbage mass over the years, the % proportions of leaves and stems had consistent relationships with sward height (Figure 2). The sward structure variables are conditioned by sward height and environmental variables (temperature, nitrogen fertilization, and light) [2,[19][20][21][22]. The highest percentage proportion due to stems was recorded in the 35 cm treatment and was the result of stem elongation ( Table 2). A minimal herbage mass of 2000 kg DM ha −1 is required for animal selection and is the recommended amount to sustain animal gains [7]. The results from this study showed that the herbage mass varied from 5230 to 11,120 kg DM ha −1 , which resulted in bulk densities of 34.8, 31.2, and 28.1 g DM m −3 for the 15, 25, and 35 cm sward height treatments. A previous study found no differences in herbage mass among the sward heights, which averaged 4000 kg DM ha −1 [19][20][21][22][23]. However, variations in soil fertility, rainfall, temperature, and solar radiation may explain the lower results recorded in this study compared to the previous study [23]. According to several studies, the potential production value for Marandu grass is about 11,000 kg DM ha −1 year −1 , which was also observed in 2013 by this study (Table 2) [14]. During the 2013 growing season, the measured herbage accumulation was 58 kg DM day −1 [19]. Therefore, based on 6 months growing season, approximately 10,000 kg of DM ha −1 is produced based on this herbage accumulation rate.
Grazing management aims to improve sward structure (physical characteristics, tissue distribution, anatomy, and leaf mass) and is an effective strategy that can be used to improve forage quality because changes in the structure variables can promote large differences in voluntary intake, which is the main factor that determines animal performance [24,25].

Nutritive Value
The observed mean CP and NDF contents found in this study varied from 14.0% to 14.7% and from 57.8% to 59.6%, respectively (Table 3). These values differed from several previous studies that used Marandu grass [10,26,27], but were in line with a study from Vietnam [28]. In the Brazilian studies, CP varied from 10% to 12% during the rainy season (November to April), and from 6% to 8% during the dry season, whereas NDF was always above 60%. A minimal herbage mass of 2000 kg DM ha −1 is required for animal selection and is the recommended amount to sustain animal gains [7]. The results from this study showed that the herbage mass varied from 5230 to 11,120 kg DM ha −1 , which resulted in bulk densities of 34.8, 31.2, and 28.1 g DM m −3 for the 15, 25, and 35 cm sward height treatments. A previous study found no differences in herbage mass among the sward heights, which averaged 4000 kg DM ha −1 [19][20][21][22][23]. However, variations in soil fertility, rainfall, temperature, and solar radiation may explain the lower results recorded in this study compared to the previous study [23]. According to several studies, the potential production value for Marandu grass is about 11,000 kg DM ha −1 year −1 , which was also observed in 2013 by this study (Table 2) [14]. During the 2013 growing season, the measured herbage accumulation was 58 kg DM day −1 [19]. Therefore, based on 6 months growing season, approximately 10,000 kg of DM ha −1 is produced based on this herbage accumulation rate.
Grazing management aims to improve sward structure (physical characteristics, tissue distribution, anatomy, and leaf mass) and is an effective strategy that can be used to improve forage quality because changes in the structure variables can promote large differences in voluntary intake, which is the main factor that determines animal performance [24,25].

Nutritive Value
The observed mean CP and NDF contents found in this study varied from 14.0% to 14.7% and from 57.8% to 59.6%, respectively (Table 3). These values differed from several previous studies that used Marandu grass [10,26,27], but were in line with a study from Vietnam [28]. In the Brazilian studies, CP varied from 10% to 12% during the rainy season (November to April), and from 6% to 8% during the dry season, whereas NDF was always above 60%.
The NDF content increased with stem proportion in the sward and as the forage matured, both of which are dependent on the environment [2,19,20]. A Pearson's correlation between NDF content and stem proportion generated a coefficient of 0.85. The increase in NDF with sward height observed in this study was probably due to variations in grass structure. Changes in plant structure alter the sward interception of photosynthetically active radiation (PAR) when the nitrogen fertilization rate is the same among treatments. The critical leaf area index concept states that when the grass sward intercepts 95% of PAR, it approaches its maximum growth rate without the plant shading itself [29]. Intercepts higher than 95% of PAR increase senescence and the stem growth rate, but decrease the grass nutritive value (decrease in CP and increase in fiber content) and leaf production [29].
The discriminatory power of variables within each principal component is measured by multiple correlations [30]. The results of principal component analysis identified that temperature and rainfall explained more than 80% of the variation in Marandu grass chemical composition (temperature explained 45% and rainfall 41%), which is in line with the findings for temperate grasses [31,32]. The simultaneous occurrence of high temperatures and drought can limit forage growth and change tissue chemical composition [32,33]. When analyzed separately, drought and temperature have been shown to alter the chemical composition of forage [33]. Elevated temperature and drought (inadequate soil moisture) change plant chemical composition by altering the concentrations of non-fibrous and fiber carbohydrates [34].

Animal Performance
In Brazil, Nellore calves are weaned at 7 months, and have an average BW of 210 kg [35]. This study used animals that were bought after weaning with initial body weights ranging from 180 to 230 kg. The effect of initial body weight on average daily gain (ADG) was attributed to the differences in the genetic group, nutrition during pregnancy, and the cow-calf phase.
Cow nutrition during pregnancy can affect the development of the progeny. Previous studies have shown that progeny from cows fed on improved pastures have higher ADGs in the finishing phase, thicker fat layers, and greater meat tenderness [36]. The development of calves from cows that were well nourished has been compared with those subjected to nutritional restrictions. The results showed that deficient nutrition can reduce the birth weight by up to 35% [37]. Furthermore, animals with lower birth weights have limited compensatory growth capabilities. In addition, when animals were slaughtered at the same age (30 months), the group with a higher birth weight showed significantly improved performances for carcass and meat cuts.
Dry matter intake (DMI) is the main variable regulating animal performance [38,39]. The DMI is controlled by numerous factors that interact with each other, such as animal characteristics, the environment, forage allowance, nutritive value, and management [40]. In this study, animal performance varied among years and sward heights.
Higher or lower ADGs and gain per hectare (GPH) values were not observed in the years with higher or lower accumulated precipitation, or the year where the grass had the highest CP (Table 1,  Table 3, and Table 5), which suggests that factors inherent to animals, such as body weight, genetic factors, and feed conversion, are mainly responsible for annual variations.
Overall ADG increased with pasture height. It has been shown that grazing selectivity changes with grazing intensity [41]. In taller swards, animals can select more leaves that have higher nutritive value, which leads to a greater ADG. Over the years, pastures with taller sward heights had greater live leaf mass per hectare and total mass per hectare values. Herbage live leaf content is a key predictor of beef cattle production because live leaves have the greatest nutritive value [10,42]. Herbage mass in this study was always greater than 5000 kg DM −1 ha −1 , which meant that it was consistently higher than the 2000 kg recommended level, below which there is a limit to forage intake in Brachiaria pastures [7].
High intakes and animal performances are achieved when high herbage masses and allowances are provided and forage nutritive value is high. Sward height affects the number of meals per animal per day when it is less than 15 cm, while the time per meal decreased with an increase in sward height [37]. Previous studies have also shown that a larger forage allowance increases the number of meals and reduces the time per meal [20,43,44]. The improved animal performance observed in this long-term study (Table 3) suggested that both nutritional and non-nutritional factors limited forage intake [39].
This study evaluated the performance of Nellore bulls in the rearing phase during the Marandu grass growing season (December to April). They received additional mineral salts and an average supplement of 150 kg N. In Central Brazil, the ADG value without supplements is only 0.3 kg animal −1 day −1 [45], which is much lower than the results recorded in this study (approximately 0.8 kg animal −1 day −1 ) (Table 3). Therefore, animal performance can be increased in tropical regions by controlling sward height. The results obtained over the 7 yrs of evaluation can be attributed to (1) greater forage mass (>2 times greater than that found to limit intake), (2) a higher proportion of live leaves in the herbage mass (>50%), (3) a grazing efficiency of 50% obtained using the put-and-take method, (4) high digestibility (>60%), (5) a high CP level for tropical grasses (>140 g kg −1 ), (6) an NDF concentration of less than 60%, and (7) an adequate protein/energy ratio (>150 g CP kg −1 DOM).
Improved grassland management regimes, where the aim is sustainable intensification, need to be based on an understanding of the seasonal variations in productivity, stocking rate, sward structure, soil fertility, and animal behavior. The focus should not only be on carcass production per animal or area. Furthermore, livestock intensification needs to reduce the environmental footprint produced by cattle farming and generate ecosystem services, such as improved soil quality, reduced erosion, and GHG emissions.

Greenhouse Gases
Enteric CH 4 is affected by the chemical composition of forage. The CH 4 amount produced by animals fed on forages rich in structural carbohydrates is greater than the amount produced by those fed diets with higher levels of non-structural carbohydrates [11,46,47]. The data from this study showed that in the first and third years, the results were in line with previous studies because they showed that NDF content and enteric CH 4 production were greater in the 35 cm pasture treatment than in the other treatments (Table 6). However, the results also showed a higher variation in the treatment responses over the evaluation years. Other variables, such as animal genetics, may explain the differences between years.
The variation in CH 4 emissions from 106 to 177 g CH 4 animal −1 day −1 among sward heights and years can be explained by dry matter intake variation (Table 6). Similar behavior was reported in previous studies where a large variation was found between daily CH 4 emissions. This variation was attributed to forage selection by the animals and, consequently, to variations in daily dry matter intake [48,49]. In the 25 cm height treatment, the NDF concentration was lower, and digestibility and the leaf appearance rate were higher, which may explain the lower enteric CH 4 production ( Table 7). Harvesting forage at an earlier grass maturity stage has often been proposed as a strategy to decrease enteric CH 4 production [50].
The IPCC guidelines for national inventories default emission factor for enteric CH 4 from beef cattle with the same range of body weights studied here are 149 g CH 4 animal −1 day −1 [51]. In this study, daily CH 4 emissions ranged from 117 to 139 g animal −1 day −1 ( Table 6). The enteric CH 4 measured in this study was lower than the average IPCC emissions factor for tropical pasture (149 g animal −1 day −1 ) but within the range of variability around the mean given in the IPCC guide. The results produced by this study appear to be the first reports of enteric CH 4 emissions for cattle that have only consumed Brachiaria grass. The amount of CH 4 emitted by tropical production systems requires measurements from long-term experiments across a specified region.
Grazing intensity effects CO 2 and N 2 O emissions [12]. Both GHG emissions declined in this study as the grazing intensity decreased. However, these effects varied according to the season, and there was no overall effect due to grazing intensity. The correlation between soil variables and GHG emissions was analyzed, and the results showed that there was a positive correlation between soil moisture and CO 2 and N 2 O, whereas soil ammonium and nitrate were not correlated with GHG emissions (Figure 1).
Soil moisture supports biological activity [52,53] and can control CO 2 and N 2 O emissions. This control is probably due to changes in root biomass, microbial decomposition, and variations in substrate supply depending on soil moisture [54]. Indeed, GHG emissions in tropical grasslands are extremely affected by seasonal variations in the rainfall regime [13,53,55]. Soil moisture also controls O 2 diffusivity and, consequently, aerobic conditions. Nitrous oxide is produced during nitrification and denitrification reactions in the soil [56]. The results produced by this study were in line with several previous studies that reported a positive correlation between soil moisture and N 2 O [13,57,58]. These studies attributed this effect to variations in N uptake by plants and microorganism populations, N transformations (mineralization, nitrification) in the soil, and, principally, increases in denitrification activity induced by reduced O 2 diffusion into the soil.
The results showed that soil NH 4 + and NO 3 + were not correlated with N 2 O and CO 2 emissions ( Figure 1). However, mineral nitrogen concentration (NH 4 + and NO 3 − ) has a controlling effect over N 2 O emissions from soils [59]. The reasons for the absence of a correlation between N 2 O production and mineral N in this study are not clear, but presumably, N was not a limiting factor during the experimental period, which suggests that it was soil moisture that drove the fluxes. Methane fluxes from soil are driven by initial C and N concentrations, soil moisture, temperature, dissolved N and C, and mineral N concentration [53,60]. The results did not show a clear correlation between CH 4 fluxes and any soil variable (Figure 1). The CH 4 emissions were probably driven by different variables depending on the season and forage management activities. It has been reported that Brachiaria pastures are not a source of CH 4 . These previous studies suggested that the well-drained conditions of this soil favored CH 4 oxidation, which consequently offset the CH 4 production [12].

Practical Implications for Pasture Management
From a production perspective, the results were consistent with the hypothesis that increasing sward height leads to a diminution in the sward leaf and crude protein content. However, due to the higher forage allowance, there is a greater possibility of animal diet selection, resulting in higher individual performance. The feed value of Marandu grass and animal performance was relatively high in the current study.
The results also showed that the lowest sward height produced the highest N 2 O emissions, which is the most potent GHG. Irrespective of pasture height, enteric CH 4 production was lower than the IPCC emissions factor for beef cattle in Latin America by approximately 30%. However, this is still within the range considered by the IPCC. The CH 4 conversion factor (Ym) factor, which represents the annual gross energy intake lost as CH 4 , was similar to that observed in high-quality forage for all sward heights. According to this IPCC guideline, high-quality forage has a Ym lower than 6% [15]. In this study, the reported Ym was approximately 5% [19].
The results showed that individual animal performance increased from 668 to 901 g day −1 , and gain per area decreased from 649 to 439 kg ha −1 during the rearing phase. The GHG emissions during the rearing phase were calculated based on the data for enteric CH 4 reported here and GHG emissions for soil reported by Cardoso et al. (2017) [12]. The GHG data were converted into equivalent carbon dioxide (CO 2 eq) using a global warming potential of 28 for CH 4 and 265 for N 2 O. The analysis showed that the carbon footprint was 3.28, 3.54, and 4.08 kg CO 2 eq kg −1 BW gain during the rearing phase for each sward treatment. A sward height of 25 cm for Marandu grass clearly produced the best forage quality and environmental performance according to the individual and per unit area gain in live weight results. However, there are some important aspects to consider before this management strategy is recommended. For example, both the soil type and water availability in this study were suitable for grazing. Furthermore, the put-and-take method was used to control the weekly stocking rate, which was based on the forage accumulation rate. In addition, nitrogen fertilizer was applied three or four times during the Marandu grass growing season. Despite this, the adoption of this management strategy by farms can result in high animal performances. Our results suggest that after seven years of grazing management that a fixed pasture height that corresponds to the 95% light interception during the growing season, which results in greater CP concentration and leaf proportion for Marandu palisade grass. Therefore, obtaining higher animal performance.

Conclusions
The principal component analysis revealed that weather-related variables had a strong effect on the structure and nutritive value of the sward. The canonical correlation analysis showed that rainfall was the main variable explaining differences in structure and nutritive value. Greenhouse gas emissions from the soil and animals were controlled by climatic variables and, to a lesser extent, by grazing intensity.
Sward height had clear effects on sward structure, nutritive value, and animal performance. Based on the data of nutritive value, animal production per area, and greenhouse gas emissions, the sward height of 25 cm is the most visible way of obtaining greater animal performance and sustainability per area using Marandu palisade grass and continuous stocking.
Further studies should quantify the interactions between N fertilization and sward height as well as on the GHG emissions from fertilized grazed grasslands. These data could form the basis for further research evaluating the potential of managing the sward height of Marandu palisade grass to achieve sustainable intensification in forage-based beef cattle production.