Relationship between residual feed intake and enteric methane emission in Nellore cattle

Feed intake and average daily gain (ADG) in Nellore cattle were determined to calculate residual feed intake in two performance tests: first during the growth phase (RFIgrowth) and then during a measurement of the methane emission phase (RFImet). During the RFIgrowth test, 62 males and 56 females were classified as low-, medium-, and high-RFI. Enteric methane emission was measured in 46 animals; 23 males used for RFImet measurement plus 23 females (22 low-RFIgrowth and 24 high-RFIgrowth). Diet consisted of Brachiaria brizantha cv. Marandu hay (445 g/kg DM) and concentrate (555 g/kg DM). During the RFIgrowth and RFImet phases, DMI was lower in the animals with low RFI, with no difference in ADG. Residual feed intake was −0.359 and 0.367 kg DM/d for lowand high-RFI animals. Enteric methane emission (g/d, g/kg BW0.75 and g/kg ADG) did not differ between RFIgrowth classes. Enteric methane emission (g/d) was higher in high RFImet and lower in low RFImet males. Spearman correlations among traits obtained during both tests, which were high between metabolic BW (r = 0.959) and between DMI (r = 0.718), and zero between ADG (r = −0.062), resulted in moderate correlation between RFIgrowth and RFImet (r = 0.412). However, it is not possible to confirm that high-efficiency animals release less enteric methane, since different results were obtained when enteric methane was compared between the RFIgrowth and RFImet classes.


Introduction
Residual feed intake (RFI) is an attractive measure of feed efficiency, as it does not depend on body size, growth rate, or body composition measures (Durunna et al., 2012).High-efficiency animals exhibit low RFI and consume less feed than expected for maintenance and production, whereas low-efficiency animals have high RFI and observed feed intake higher than the predicted feed intake.
The possibility of improving production efficiency by exploiting the genetic variation in RFI and possible variation in traits related to emission of greenhouse gases is of great economic importance for beef cattle production.Studies have shown that selection of animals for low RFI (high feed efficiency) reduces methane emissions in cattle (Nkrumah et al., 2006;Hegarty et al., 2007;Fitzsimons et al., 2013).However, there are inconsistencies in relation to differences in feed intake and corresponding methane emissions between low-and high-RFI animals.Fitzsimons et al. (2013) offered grass silage to heifers and observed that both DMI and methane emission (g/d) were lower for low-RFI compared with high-RFI animals, and Jones et al. (2011) reported that DMI was similar between divergent RFI groups and that methane emission was similar when animals grazed low-quality pastures but were lower for low-RFI compared with high-RFI animals when grazing highquality pastures.The conflicting observations regarding the variation in methane emission of high-and low-efficiency animals between studies suggest that selection of cattle for increasing feed efficiency does not necessarily result in a reduction in methane emission (Freetly and Brown-Brandl, 2013).To date, published studies relating RFI and methane emission are limited to Bos taurus cattle, and little information is available on Bos indicus cattle.
In view of these considerations, the objective of the present study was to evaluate the relationship between enteric methane emission and residual feed intake in Nellore cattle fed a high-roughage diet.

Material and Methods
The experiment was approved by the Animal Ethics Committee of Instituto de Zootecnia, Nova Odessa, Brazil, in accordance with Guidelines of Animal Welfare and Humane Slaughter (São Paulo State law number 11.977).
The study was conducted in Sertãozinho, northern region of the state of São Paulo, Brazil (21°10' South latitude and 48°5' West longitude).The climate of the region is tropical humid, with average annual temperature and rainfall of 24 °C and 1,312 mm, respectively.
Animals were housed in individual pens: one facility destined for males and two other facilities for females to fit the number of evaluated animals.Voluntary feed intake was calculated as the difference between feed supplied and refusals.The amount of feed offered was adjusted daily based on the intake observed on the previous day to maintain refusals at about 10% of the feed supplied.Diet samples were collected at intervals of 28 d and the refusals per animal were weighed daily and sampled twice a week.The samples were stored in properly labeled plastic bags and frozen.
Samples of feed and refusals were oven-dried at 55 °C for 72 h and ground through a 1-mm screen (Wiley mill, Arthur Hill Thomas Co., Philadelphia, PA).These samples were analyzed for nitrogen (N) (AOAC Official Method 984.13), ash (AOAC Official Method 942.05), and EE (AOAC Official Method 920.39) contents, in accordance with the AOAC (1990).Neutral (NDF) and acid (ADFom) detergent fiber analyses were based on procedures described by Mertens (2002).Lignin (sa) was estimated as described by Van Soest and Robertson (1985).Gross energy was determined using an adiabatic bomb calorimeter (model 6300, Parr Instrument Company, Moline, IL, USA).
Two performance tests (males and females) were conducted to calculate RFI growth .For males, the performance test lasted 107 d, whereas for females it lasted 112 d (Figure 1).The adaptation periods were 35 and 28 d for males and females, respectively, and these data were not included in the RFI growth period.Therefore, serial BW and individual feed intake data were obtained for 72 (July 5 to September 14, 2011) and 83 d (August 3 to October 25, 2011) for males and females, respectively.The males' ADG was determined based on two records per week on consecutive days at intervals of 14 d without fasting (total of 22 weight records).For females, individual BW records were obtained at intervals of 28 d, with deprivation of water and feed for 16 h (total of four weight records).RFI growth -residual feed intake in the growth phase; RFI met -residual feed intake in the methane emission phase.
Figure 1 -Chronological plan for RFI growth (males and females), RFI met (males), and measurement of enteric CH 4 (males and females) in the experimental period.
After the RFI growth period, 23 males were sampled for RFI met analysis -11 lowest RFI growth and 12 highest RFI growth .These animals continued for an additional 45 d in individual pens (including the period of methane collection) with daily records of DMI.However, to meet the herd selection criteria (Mercadante et al., 2003), one fasting weighing was performed on d 9, causing the first 13 d to be invalidated.Therefore, the performance test was conducted over the last 32 d (September 27 to October 30, 2011) (Figure 1).Average daily gain was determined based on nine weighing records of animals without previous fasting, which allowed for the calculation of a new RFI (RFI met ), more representative of the methane-emission measurement period.This was done despite the fact that at least 63 d of evaluation are necessary to obtain accurate estimates of ADG (Wang et al., 2006).
Enteric methane emission (g/d) was measured in 46 animals (Figure 2) -22 low-RFI growth animals (11 males and 11 females) and 24 high-RFI growth animals (12 males and 12 females).Data were collected for six consecutive days over a period of 24 h, from the 148th to the 153th day for males (October 25 to October 30, 2011) and from the 175th to the 181th day for females (November 21 to November 27, 2011) (Figure 1), using the sulfur hexafluoride (SF 6 ) tracer gas technique as described by Johnson and Johnson (1995), adapted by Primavesi et al. (2004).The permeation tubes were filled with approximately 1.0 g SF 6 .To confirm the emission rate of SF 6 , the permeation tubes were kept in a water bath at 39 °C and weighed eight times to construct an 8-point regression curve.The tubes were administered orally to each animal in a random manner 4 d before the beginning of methane sampling.
Expired and eructated gases, together with ambient air, were sampled in 60-mm class 20 polyvinyl chloride (PVC) canisters through a stainless steel capillary tube (inner diameter: 0.127 mm) fixed to the halter of the animals.The sampling canisters were evacuated before placement into the animals and gases were analyzed about 5 d after sample collection, as the experiment was conducted far from the laboratory of gas analysis.
To correct atmospheric concentrations of methane inside the facility, gas samples were collected from ambient air with two sampling canisters per day ("blank"), hung at the entrance and exit of the facility.At the end of the sampling period, SF 6 and CH 4 concentrations in the samples were determined by gas chromatography using an HP6890 gas chromatograph equipped with a flame ionization detector (FID) and Plot HP-Al/M megabore column (0.53 μm, 30 m) for CH 4 , and with an electron capture detector (μ-ECD) and HP-MolSiv megabore column for SF 6 , using two 0.5-cm 3 loops coupled to two 6-way valves.The canisters were pressurized with nitrogen 5.0 (White Martins, Praxair Inc) until reaching a pressure of approximately 1.2 atm.Pressure readings were obtained with a digital pressure meter.The calibration curves were constructed, using gas standards certified by White Martins (Praxair Inc), as described by Westberg et al. (1998).The flow rate of CH 4 released by the animal was calculated in relation to the flow of SF 6 , correlating the results with the known release rate of SF 6 in the rumen, subtracted from baseline CH 4 concentrations (Westberg et al., 1998) Blaxter and Clapperton (1965), corrected by Wilkerson et al. (1995), was used to calculate the loss of gross energy in the form of methane: CH 4 /GEI = [(CH 4 × 0.0133)/GEI] × 100, in which GEI is the gross energy intake and 0.0133 is the gross energy in Mcal/g CH 4 (Holter and Young, 1992).
Residual feed intake was calculated as the difference between observed DMI and DMI predicted by regressing DMI on ADG and average BW 0.75 (Koch et al., 1963).The dry matter intake was calculated based on the mean of feed intake values obtained during each test period (RFI growth and RFI met ) multiplied by the dry matter concentration of the diet.The average daily gain in each test was estimated as the linear regression coefficient of BW on j days of test:  to ADG; Dot i = days of test for the i-th observation; and ε i = random error associated with each observation.The mean BW 0.75 during the test was calculated as: BW 0.75 = [α + β × (Dot / 2)] 0.75 , with α and β assuming the values obtained by the equation described above (Grion et al., 2014).
Residual feed intake was estimated using the following regression equation: DMI = β T × TG + β TA × TG × ADG + β TB × TG × BW 0.75 + ε (i.e., RFI), in which β T , β TA , and β TB are regression coefficients of the classificatory variable test group (TG) and of the interactions between TG and the covariates ADG and BW 0.75 , respectively; and ε is the residual of the equation (i.e., RFI).TG is the test group [sex and two facilities only for females, TG = 1 (male, n = 62), 2 (female, n = 31), and 3 (female, n = 25)].Animals were classified into three groups: low RFI growth (<mean -0.5 SD; n = 40); medium RFI growth (± 0.5 SD of the mean; n = 42), and high RFI growth (>mean + 0.5 SD; n = 36), taking into consideration the standard deviation (SD) of the mean RFI (SD = 0.309 kg DMI for males and SD = 0.329 kg DMI for females).For RFI met , males were classified as low-RFI (<zero) and high-RFI (>zero), considering that the mean RFI is zero.
Variables were compared between two RFI classes (low and high) by the least square method.The statistical model included the effects of RFI growth class (low or high RFI) and sex (male or female), or the effect of RFI met class.The interaction between RFI growth class and sex was not significant for the analyzed variables and was therefore removed from the model.Spearman correlations were computed to compare ADG, DMI, BW 0.75 , and RFI across the two test periods for the male calves.

Results
No significant differences in initial BW, final BW, BW 0.75 , or ADG were observed between low-and high-RFI growth animals (Table 2).RFI growth was −0.359 and 0.367 kg DM/d for low-and high-RFI growth animals, respectively, corresponding to a difference of 0.720 kg/d in DMI between high and low-efficiency animals.Animals classified as high-RFI growth consumed 11% more than low-RFI growth animals.Enteric methane emission (g/d and g/kg BW 0.75 ) did not differ when compared between RFI growth classes (P>0.05).On the other hand, methane emission expressed as g/kg DMI was higher in low-RFI growth animals (P<0.001).No effect on enteric methane emission was found in RFI growth when expressed as g/kg ADG (P = 0.304) (Table 2).
After the RFI met test period (32 d), males were re-ranked as low-and high-RFI met (n = 23).The animals did not differ in terms of BW 0.75 or ADG (P>0.05) between RFI met classes.However, DMI was higher in bulls classified as high-RFI met (8.12±0.14kg/d) compared with the low-RFI met (7.09±0.17kg/d).The same was observed for gross energy intake (GEI), which was 33.5±0.55Mcal/d and 29.5±0.69Mcal/d for highand low-RFI met animals, respectively.Results of −0.506 and 0.362 kg DM/d were found for animals with low and high RFI met , respectively, with a variation of 0.868 kg/d in DMI between high and low-efficiency animals, corresponding to a 13% higher intake in high RFI met animals (Table 3).When methane emission (g/d) was compared between RFI met classes, it was higher for high RFI met than low RFI met animals (P<0.05).Methane emission (g/kg DMI and %GEI) did not differ between RFI met classes (P>0.05).
A positive correlation was observed for BW 0.75 (r = 0.959) and DMI (r = 0.718) between RFI growth and RFI met , even considering the short duration of the RFI met test (32 d).However, a correlation close to zero was observed for ADG (r = −0.062) between test periods.Spearman correlation showed a positive association between RFI growth and RFI met (r = 0.412) (Table 4), with strong evidence of animal reranking between RFI growth and RFI met .
77% for heifers fed corn silage and a concentrate diet (Kelly et al., 2010).Apparently, in the present study, the coefficient of determination of the model used for the estimation of RFI is not associated with the dietary energy density, as reported by Fitzsimons et al. (2013).A lower coefficient of determination (R 2 = 0.54) was obtained for the estimation of RFI met when the same diet was offered, explaining 54% of the variation in DMI.This value was higher than the 43% observed by Lancaster et al. (2009) for heifers fed alfalfabased diet and concentrate, and the 38% observed by Freetly and Brown-Brandl (2013) for heifers fed a high-roughage diet.The short duration (32 d) of ADG measurement may have compromised the calculation of RFI met .In that equation, ADG was not significant (P = 0.07) and BW 0.75 explained most of the variation in DMI.
When RFI is used as a measure of feed efficiency, efficient animals consume less than their inefficient counterparts for the same production performance (Fitzsimons et al., 2013).As a consequence, it would be expected that animals with low RFI would release less methane proportionally to their lower feed intake.However, studies have shown variations in methane emission between low-and high-efficiency animals.Fitzsimons et al. (2013) observed less methane emission (g/d and g/kg BW 0.75 ) in low-RFI heifers and Hegarty et al. (2007) showed that steers selected for low RFI also have reduced daily methane emission (g/d).These results agree with the observations in the RFI met test, in which Nellore bulls classified as low-RFI met released less enteric methane (g/d) (P = 0.024).Taken together, the results suggest the potential selection of cattle for feed efficiency and reduced enteric methane emission.On the other hand, no difference (P>0.05) in enteric methane emission (g/d and g/kg BW 0.75 ) was observed in the present study when it was compared between animals of low and high RFI growth .
The methane emission data (g/d) are similar to those reported by Possenti et al. (2008), who observed emission of 147.5 g CH 4 /d in cattle fed coast-cross hay with 200 g/kg of leucaena, and lower than those estimated by Johnson and Johnson (1995) for beef cattle (164 to 194 g CH 4 /d).The values expressed as kg/year (51-53 kg CH 4 /kg year) agree with the estimates reported by Cederberg et al. (2009) for young cattle in Brazil (47-56 kg CH 4 /year), and by the IPCC (2006), which estimated an average emission of 49 kg CH 4 /year for young cattle (230 kg BW) in Latin America.These data corroborate the findings of Jones et al. (2011), who observed similar daily methane emissions (g/kg BW) in cows with low and high RFI grazing on low-quality pasture.
When methane emissions were expressed as unit of feed consumed (g/kg DMI during RFI growth ), high-efficiency animals (low-RFI growth ) released more methane (25.1 g/kg

Discussion
No differences in ADG were observed between RFI growth classes (Table 2), despite a difference in DMI between low-and high-RFI growth animals (P>0.001)due to the definition of RFI itself (Koch et al., 1963).This fact has also been reported in recent studies demonstrating that RFI is independent of ADG in cattle (Fitzsimons et al., 2013;Durunna et al., 2012;Lawrence et al., 2012;Cruz et al., 2010).The ADG of animals classified as low-and high-RFI (0.754 vs 0.744 kg/d, P = 0.810) during the RFI growth test was moderate and is consistent with that observed for weaned Nellore animals fed Brachiaria brizantha hay-based diet and concentrate.Similar ADG values (0.820 kg/d) were reported by Sobrinho et al. (2011) for Nellore bulls selected for post-weaning weight that received a similar diet.On the other hand, ADG was numerically lower in the animals classified as low-and high-RFI (0.466 and 0.547 kg/d, respectively; P = 0.726) during the RFI met test when the animals received the same diet.Even though the diet and facility were the same and the DMI expressed as percentage of average test BW (2.2%) were similar in growth and methane phases, lower ADG was observed during RFI met compared with RFI growth , probably due to the daily management for CH 4 emission measurement.
The linear regression model used to calculate RFI growth explained 82% of the variation in DMI.This value was higher than those reported in the literature, with a variation of 60% for heifers fed grass silage (Fitzsimons et al., 2013); 66% for heifers fed grass silage and concentrate as supplement (Lawrence et al., 2012); 70% for crossbred steers receiving a high-corn diet (Freetly and Brown-Brandl, 2013); and DMI) than their low-efficiency counterparts (22.8 g/kg DMI) (P<0.05).In this case, the enteric methane emission is probably related to higher digestibility and a higher ruminal fermentation rate, increasing the total production of short-chain fatty acids with a proportional increase in the production of methane.Previous research (Magnani et al., 2013) reported higher DM digestibility for low-RFI Nellore heifers than high-RFI heifers (49.14% vs 45.38%, respectively).Richardson et al. (1996) observed that steers and heifers with low or high feed efficiency tend to differ in their capacity to digest DM, by about one percentage unit (DM digestibility of 68%), and this difference in digestibility was responsible for 14% of the difference in DMI.However, digestibility and ruminal digestion products were not evaluated in the present study, but studies suggest that the efficiency of animals classified as low-RFI is associated with a greater digestive capacity (Herd and Arthur, 2009;Richardson and Herd, 2004).Freetly and Brown-Brandl (2013) suggested that methane emission may increase with increasing feed efficiency and that selection of cattle for increased feed efficiency will not necessarily reduce enteric methane emission.
The results reported by Freetly and Brown-Brandl (2013) also do not support the hypothesis that increased feed efficiency reduces methane emission.The authors observed that measures of feed efficiency, BW gain:DMI ratio, and RFI, are not associated with methane emission in cattle fed a high-corn diet (828 g/kg).When the animals were fed a high-roughage diet (600 g/kg of corn silage, 30 g/kg of alfalfa hay), RFI was not related to methane emission, but the association between BW gain:DMI ratio and methane emission was positive.According to the authors, the increase in the methane emission rate with increasing feed efficiency (BW gain:DMI) may have been due to the higher fermentation of feed, increasing nutrient availability and methane emission.In contrast, no differences in methane emission (g/kg DMI) were observed between animals classified as low-and high-RFI met (P = 0.915), despite differences in DMI between low-and high-RFI met animals (7.09 and 8.12 kg/d, respectively) and a moderate positive Pearson correlation (r = 0.62) between CH 4 (g/d) and DMI (kg/d).Likewise, regression analysis indicated that an increase of 1 kg in DMI was associated with an increase of 17 g/d in methane emission (r = 0.38).Fitzsimons et al. (2013) and Hegarty et al. (2007) also observed no difference in methane produced as a function of DMI between RFI classes; however, Nkrumah et al. (2006) observed that low-RFI steers released less methane per unit of DMI.
The methane emission values (expressed as g/kg DMI) estimated in the present study (20.1 to 25.1 g/kg DMI) agrees with those reported in the literature for animals fed high-roughage diets.McGeough et al. (2010) observed methane emissions of 35.6 g/kg DMI in beef cattle fed grass silage plus 2.60 kg of supplemental concentrate DM, and 15.3 g/kg DMI for animals offered concentrate ad libitum.These results suggest that animals fed high-forage diets generally tend to release more methane (g/kg DMI) than animals offered high-concentrate diets (Janssen, 2010).In the present study, no difference in methane emissions, expressed as g/kg ADG during RFI growth or RFI met , was observed between more and less efficient animals (P>0.05).This finding was expected, since no significant difference in ADG was observed during both studied periods (P>0.05).
The loss of energy in methane form, defined as the percentage of enteric methane emitted and expressed as a function of gross energy intake (%GEI), was on average 6.5% during the RFI met period.This value agrees with those predicted by the IPCC (2006) for beef cattle raised under tropical conditions (6.5% to 7.5%).However, these values are lower than those reported by McGeough et al. (2010) for continental crossbred steers fed corn silage-based diets (7.3% to 8.4% GEI) using SF 6 gas tracer techniques, and by Fitzsimons et al. (2013) for grass silage-based diets offered to Simmental heifers (12.6% GEI), with values established at the upper end of variation of 2% to 12% GEI as reported by Johnson and Johnson (1995).
High-efficiency animals show a significant reduction in DMI; however, it is not possible to confirm that more efficient animals release less enteric methane, since different results were obtained when enteric methane was compared between RFI growth classes and RFI met classes.These results may be expected, since the duration of the tests differed, as also did the number of days between each test and methane measurement.While RFI growth was based on measurements obtained in the middle of the test, about 65 days before methane sampling, RFI met was measured exactly during the measurement of methane emission.In addition, the period of 32 d to obtain RFI met may not be sufficient for the accurate measurement of ADG (Wang et al., 2006).This would explain the Spearman correlations between traits obtained during both tests, which were high between BW 0.75 and DMI and zero between ADG, resulting in a correlation of medium magnitude between RFI growth and RFI met , with evidence of re-ranking of the animals for RFI.
In view of the differences in enteric methane emission among studies, further research is needed to determine the effect of selection of high-efficiency animals (low RFI) on enteric methane emission by investigating different diets and phases of the animal lifetime and methods used for measuring enteric methane.Several mitigation strategies have been proposed to reduce greenhouse gas emission, many of them designed to reduce enteric methane emission.The selection of animals with low DMI and similar ADG could be a strategy to reduce greenhouse gases emissions in the beef production system as a whole since these animals are highly efficient in feed utilization.

Conclusions
There is no evidence that high-efficiency animals release less enteric methane, even with lower dry matter intake and same performance than their inefficient counterpart.
RFI identification, Nellore cattle born in 2010 were evaluated in individual performance tests, in an incomplete block design.One hundred eighteen animals, consisting of 62 males (starting at 261±23.1 d of age and 243±40.1 kg of BW) and 56 females (starting at 290±26.3 d of age and 216±30.8kg BW), were evaluated for RFI growth to identify high-(low RFI), intermediate-(medium RFI) and lowefficiency (high RFI) animals.After RFI growth identification, enteric methane emission measurement was performed in 46 animals (23 males and 23 females).Next, in order to obtain an estimate of individual feed intake and RFI at the time of enteric methane emission measurement, a new RFI (RFI met ) was calculated, but only in males (n = 23; starting at 351 d of age and 343 kg BW).
emission rate of CH 4 by the animal; Q SF6 = known release rate of SF6; [CH 4 ] Y = concentration of CH 4 in the canister; [CH 4 ] B = concentration of CH 4 in the blank; and [SF 6 ] = concentration of sulfur hexafluoride in the canister.The equation of in which y i = BW in the i-th observation; α = intercept of the regression equation corresponding to the initial BW; β = linear regression coefficient corresponding Open symbols represent all animals tested for residual feed intake during the growth phase (RFI growth ; n = 118) and black solid symbols represent animals used for measurement of enteric methane emission (n = 46).

Table 1 -
Nutrient composition of the ingredients and total diet offered to the animals during the period of evaluation of residual feed intake (Weiss, 1999)s percentage of dry matter (g/kg DM). 2 Values calculated by Weiss' equation(Weiss, 1999).3Valuescalculated by NRC (2000) level 2.

Table 2 -
Performance traits, enteric methane emission, and feed efficiency of Nellore males and females classified as low-and high-residual feed intake (RFI growth ) BW -body weight; BW 0.75 -metabolic body weight; DMI -dry matter intake; ADG -average daily gain; RFI -residual feed intake.

Table 3 -
Performance traits, enteric methane emission, and feed efficiency of young Nellore males classified as lowand high-residual feed intake (RFI met ) BW -body weight; BW 0.75 -metabolic body weight; DMI -dry matter intake; GEI -gross energy intake; ADG -average daily gain; RFI -residual feed intake. 1 Same records were shown in Table2, excluding females.

Table 4 -
Spearman correlation coefficients between performance traits and residual feed intake (RFI) of Nellore cattle obtained during the two evaluation periods of RFI (RFI growth and RFI met )