Phenotypic association among performance, feed efficiency and methane emission traits in Nellore cattle

Enteric methane (CH4) emissions are a natural process in ruminants and can result in up to 12% of energy losses. Hence, decreasing enteric CH4 production constitutes an important step towards improving the feed efficiency of Brazilian cattle herds. The aim of this study was to evaluate the relationship between performance, residual feed intake (RFI), and enteric CH4 emission in growing Nellore cattle (Bos indicus). Performance, RFI and CH4 emission data were obtained from 489 animals participating in selection programs (mid-test age and body weight: 414±159 days and 356±135 kg, respectively) that were evaluated in 12 performance tests carried out in individual pens (n = 95) or collective paddocks (n = 394) equipped with electronic feed bunks. The sulfur hexafluoride tracer gas technique was used to measure daily CH4 emissions. The following variables were estimated: CH4 emission rate (g/day), residual methane emission and emission expressed per mid-test body weight, metabolic body weight, dry matter intake (CH4/DMI), average daily gain, and ingested gross energy (CH4/GE). Animals classified as negative RFI (RFI<0), i.e., more efficient animals, consumed less dry matter (P <0.0001) and emitted less g CH4/day (P = 0.0022) than positive RFI animals (RFI>0). Nonetheless, more efficient animals emitted more CH4/DMI and CH4/GE (P < 0.0001), suggesting that the difference in daily intake between animals is a determinant factor for the difference in daily enteric CH4 emissions. In addition, animals classified as negative RFI emitted less CH4 per kg mid-test weight and metabolic weight (P = 0.0096 and P = 0.0033, respectively), i.e., most efficient animals could emit less CH4 per kg of carcass. In conclusion, more efficient animals produced less methane when expressed as g/day and per kg mid-test weight than less efficient animals, suggesting lower emissions per kg of carcass produced. However, it is not possible to state that feed efficiency has a direct effect on enteric CH4 emissions since emissions per kg of consumed dry matter and the percentage of gross energy lost as CH4 are higher for more efficient animals.


Introduction
Enteric methane (CH 4 ) emission is a natural process in ruminants that can result in losses of 2 to 12% of the total energy consumed by the animal [1]. The variation is the result of some factors such as chemical composition of the diet, intake level [2], and even genetic [3] and metagenomic [4].
Residual feed intake (RFI) has been used as a selection criterion in beef cattle in order to increase individual feed efficiency [5,6]. Most efficient animals (negative RFI) have a significant economic advantage as they consume less dry matter than expected for their weight and weight gain [7]. Consequently, the use of negative RFI animals has the potential to significantly reduce meat production costs.
Generally, the higher the dry matter intake (DMI), the higher the daily enteric emissions of CH 4 since a larger amount of substrate will be available for fermentation in the rumen and consequently more hydrogen will be available for methanogenesis [8,9]. Therefore, the use of more efficient animals may reduce enteric CH 4 emissions proportionally to the lower feed intake [10]. However, it is unclear whether the differences in enteric CH 4 emissions are due to the variation in digestive efficiency between negative and positive RFI animals or simply the result of from the lower DMI associated with negative RFI animals [11,12].
Inconsistencies still exist regarding the relationship between feed efficiency (RFI and FC) and enteric CH 4 emission by cattle. Studies have shown this correlation is positive and favorable in the case of highly digestible diets [10,13,14], while the phenotypic relationship between feed efficiency and enteric CH 4 emission is zero or even negative and unfavorable in diets with low digestibility [14][15][16][17], suggesting that individual enteric CH 4 emission may even increase with the improvement of feed efficiency. Furthermore, few studies have investigated Bos indicus animals receiving high roughage diets [16,18,19] and there is a lack of studies involving a large number of zebuine animals. The aim of the present study was to evaluate the relationship among performance, feed efficiency and enteric CH 4 emission traits in Nellore cattle (Bos indicus). The hypothesis was that the use of animals with low DMI and similar ADG could be a strategy to reduce greenhouse gases emissions in the beef production system.

Location and animals
The data were collected in 2011, 2012, 2018, 2019 and 2020 in Sertãozinho-SP, Brazil, as well as in Botucatu-SP, Brazil, in 2019. Performance, feed efficiency and enteric CH 4 emission data were obtained from 489 Nellore animals evaluated in performance tests. This study was carried out in strict accordance with the recommendations in the Guidelines for Animal Welfare and Humane Slaughter (São Paulo State, Law Number 11.977). The protocol was approved by the Committee on the Ethics of Animal Experiments of the Institute of Animal Science (Protocol Number 278-19), Nova Odessa-SP, Brazil.

Treatments and management
The performance tests had an average duration of 76.5 ± 12 days preceded by 28 days of adaptation [5]. The animals started the test at 376 ± 164 days of age, from June to December of each year, and were kept in individual pens (n = 95) or collective paddocks equipped with electronic feed bunks (GrowSafe1, Airdrie-AB, Canada; or Intergado1, Contagem-MG, Brazil) for automated recording of individual daily feed intake (n = 394), with ad libitum access to diet and water. The animals were weighed at the beginning and end of the test after fasting for 14 h, or at predetermined intervals without previous fasting (Table 1).
In each test, the animals were fed a single diet that differed among the years. The diets during the performance tests consisted of silage (corn or sorghum), Brachiaria hay, sugar cane bagasse, meal (cottonseed, soybean or peanut), corn (ground or wet grain), citrus pulp, mineral premix, salt, ammonium sulfate, and urea (Table 2).
After pre-drying (55 ± 5˚C for 72 h), diet samples were ground in a Willey-type mill (R-TE-650 model, Tecnal Equipamentos Científicos, Piracicaba, São Paulo, Brazil) to pass a 1-mm screen and analyzed for dry matter (method 934.01), ash (method 942.05) and ether extract (method 920.39) contents following the AOAC [22] guidelines. The contents of neutral detergent fiber (NDF) and acid detergent fiber (ADF) were determined by the methodology of Mertens [23] using a Tecnal fiber analyzer (TE-149, Piracicaba, São Paulo, Brazil) using α-amylase and without sodium sulphite. The NDF and ADF were expressed exclusive of ash. The determination of crude protein (method 990.03) was performed by the Dumas method [24] based on the release of nitrogen by combustion at high temperature in pure oxygen in DUMATHERM1 analyzer. Total carbohydrates were calculated according to the methodology described by Sniffen et al. [25]: CHOT = 100 -(CP + EE + MM); and non-fiber carbohydrates were obtained by subtracting the NDF. The gross energy (GE) determinations were performed in an adiabatic calorimetric pump of the brand IKA WERKE Model C5003 (Parr Instrument Company, Illinois, USA).
The amount of feed was adjusted weekly to guarantee daily leftovers of 5 to 10% of the total amount supplied in order to ensure ad libitum intake. The troughs were cleaned and leftovers were removed and discarded three times per week. Intake records were discarded when there were no feed leftovers and in the case of evidence of malfunctioning of the electronic measurement devices. Weekly samples of the ingredients were obtained for determination of the dry matter content of the diet.

Animal performance and feed efficiency
The following traits were calculated as described by Grion et al. [5] and Ceacero et al. [6]: DMI, average daily gain (ADG), metabolic body weight (BW 0.75 ), RFI, and feed conversion (FC). The DMI was obtained as the mean of all valid days during the period. The ADG was estimated by the linear regression coefficient of weights on days in test (DIT) according to the equation: yi = α + β x DITi + εi, where yi = weight of the animal in the i th observation; α = intercept representing the initial weight of the animal; β = linear regression coefficient representing ADG; DITi = days in test in the i th observation; εi = random error associated with each observation. The BW 0.75 was obtained as follows: BW 0.75 = (BWi + (0.5 DIT x ADG)) 0.75 , where BWi = initial body weight and DIT = days in test.
The RFI was calculated as the difference between observed and expected DMI, which was estimated by multiple regression of DMI on ADG and BW 0.75 within the test group [i = 1,. . ., 12; formed by year of birth, sex (48 females and 441 intact males), facility, site)], using the GLM procedure (SAS Inst., Inc., Cary, NC). The FC was obtained as the ratio between DMI and ADG. Mean residual gain was calculated as the difference between observed and expected ADG, which was estimated by multiple regression of ADG on DMI and BW 0.75 within the test group, using the same procedure as described above.

Ruminal methane measurement
The modified sulfur hexafluoride (SF 6 ) tracer gas technique described by Deighton et al. [26] was used for methane collection. The technique uses a permeation tube or capsule administered to the animal and deposited in the rumen. These capsules were calibrated and prepared specifically for each sampling year. The mean SF 6 gas emission of the capsules used during each sampling period was similar, with minimal variation in mg/day (Table 1). Before the beginning of the sampling period, the animals were adapted to the sampling apparatus for at least seven days. Methane gas was collected for five consecutive days, with the evacuated sampling polyvinyl chloride canisters (n = 95 animals) [16,18] or stainless-steel cylinder (n = 394 animals) being changed every 24 hours (Fig 1). The gas expelled through the mouth and nostrils of the animal was aspirated under vacuum with a capillary tube fixed in a halter and connected to the collector container, which was attached to the neck of the animal (polyvinyl chloride canister) or to a saddle on the back of the animal (stainless-steel cylinder) ( Table 1). Collector tubes were kept in the same environment as the animals to measure background concentrations of CH 4 and SF 6 during the sampling period. After each sampling period, the collectors were sent for gas chromatography analysis and their content was diluted with pure nitrogen to determine the quantities of SF 6 and CH 4 gases. The background concentrations of CH 4 and SF 6 measured by chromatography were subtracted from the concentrations found in the evacuated sampling containers of the animals.
There was a total of 12 CH 4 sampling periods (Table 1). Of the 489 animals evaluated, samples from 481 animals could be used. The losses were due to problems with the capsules. The sampling periods were September 2011; October and December 2012; November and December 2018; June, August and October 2019, and August 2020.
A gas chromatograph (HP6890, Agilent, Wilmington, Delaware, USA) was used for the analysis of CH 4 (ppm, parts per million) and SF 6 (ppt, parts per trillion). The concentrations of CH 4 and SF 6 collected in the evacuated sampling containers were determined with a flame ionization detector at 280˚C (HP-Plot Al 2 O 3 M column, 30 m length × 0.53 mm i.d. × 15 μm film thickness) and an electron capture detector at 300˚C (HP-Plot MoleSieve column, 30 m length × 0.53 mm i.d. × 25 μm film thickness), respectively, with two loops of 0.5 cm 3 maintained at 80˚C attached to 2 six-way valves. Chromatography analysis was carried out immediately after the end of the field sampling periods, which allowed the reuse of the evacuated containers in the subsequent sampling period.

Methane-related variables
Daily CH 4 emission (g/day) of each animal was obtained as the arithmetic mean of emissions on five consecutive sampling days. Enteric methane emission was also expressed as: CH 4 emission expressed per DMI (CH 4 /DMI, g/kg), ADG (CH 4 /ADG, g/kg), mid-test body weight (CH 4 /MBW, g/kg) and BW 0.75 (CH 4 /BW 0.75 , g/kg), residual CH 4 emission (observed CH 4predicted CH 4 by regression of CH 4 on DMI as described by Donoghue et al. [3]), and CH 4 emission expressed per gross energy intake (CH 4 Mcal/100 Mcal GE, as described by IPCC [27]).

Statistical analysis
The animals were classified as negative RFI (RFI<0) or positive RFI (RFI>0). The variables were analyzed using the MIXED procedure (SAS Inst., Inc., Cary, NC), fitting a model that included the fixed effect of RFI class (i = 1, 2), age of animal at the start of the performance test as covariate (linear effects), and the random effects of test group (i = 1,. . ., 12), in addition to the residual random effect. The relationships of CH 4 (g/day) with DMI, ADG and MBW were explored by Pearson's correlation and regression analyses using the CORR and GML procedures (SAS Inst., Inc., Cary, NC). The regression model for CH 4 (g/day) included the linear effect of DMI or ADG or MBW as covariate and the random effects of test group and residual. Statistical significance was declared when P<0.05.

Results
The mean weights (initial, mid-test and metabolic) or ADG did not differ between animals classified as negative and positive RFI ( Table 3). The mean RFI was -0.556 and 0.565 kg DM/ day for negative and positive RFI animals, respectively, showing a difference in DMI of 1.16 kg/day between the most and least efficient animals. Animals classified as negative RFI

PLOS ONE
Performance, feed efficiency and methane emission traits in cattle consumed on average 13% less DM than animals classified as positive RFI; consequently, FC and residual ADG higher for more efficient animals. There was a 5% reduction of CH 4 emission (g/day) in negative RFI animals compared to animals with positive RFI. In addition, despite a similar performance, more efficient animals emitted less methane expressed as g CH 4 /kg MBW and g CH 4 /kg BW 0.75 . Conversely, lower CH 4 emission in relation to DMI (g CH 4 /kg DMI), lower residual CH 4 emission and a lower percentage of GE lost as CH 4 were observed in positive RFI animals compared to animals with negative RFI ( Table 3). The simple correlation coefficients of CH 4 (g/day) with DMI, ADG and MBW were 0.77, 0.70, and 0.78 (P<0.0001), respectively. Scatter plots of CH 4 (g/day) with DMI, ADG and MBW and their respective regression equations are shown in Figs 2-4. For each kg of DMI the animals emitted on average 17.5 g CH 4 /day, for each kg of ADG the animals emitted on average 58.0 g CH 4 /day, and for each kg of MBW the animals emitted 36.0 g CH 4 /day. The regression equations of CH 4 on DMI, ADG and MBW within RFI class differed from one another (P = 0.001), accompanying the results shown in Table 3.

Discussion
In general, studies investigating the relationship between feed efficiency and enteric CH 4 emissions in cattle did not include a large number of animals, mainly because of the difficulty in measuring individual enteric CH 4 emissions in the animals, and the ones that did were conducted on Bos taurus (Table 4). In contrast, the present study evaluated 489 Bos indicus animals and enteric CH 4 was measured individually by the SF 6 tracer gas technique.
Greater reductions in enteric CH 4 emissions (15-30%) were reported in the literature for more efficient taurine animals [10,13,32] compared to the reduction of approximately 5% in the emission of zebuine animals classified as negative RFI in the present study (Table 3).

PLOS ONE
Performance, feed efficiency and methane emission traits in cattle Results similar to those of the present study were observed in crossbred taurine heifers and cows classified as more efficient, with a daily CH 4 reduction of 2.5% and 3.7%, respectively, compared to less efficient animals [34]. Lower enteric CH 4 emissions were also reported for negative RFI Angus cows compared to positive RFI cows grazing on high-quality pasture [14]. However, there was no difference in CH 4 emissions (g/day) for animals grazing a pasture of low nutritional quality [14]. On the other hand, Freetly and Brown-Brandl [15], Velazco et al. [17], Flay et al.
[33] and Batalha et al. [19] found no differences in CH 4 emissions (g/day) between more and less efficient animals.
The difference in the DMI of the animals might be responsible for the differences in daily enteric CH 4 emissions between RFI classes [9,35], which would explain the results found in the present study ( Table 3). Given that they have the same body weight, same weight gain and same amount of body fat, negative RFI animals tend to emit less CH 4 per day because of lower DMI [34]. Some studies evaluating CH 4 emission in animals classified as negative and positive RFI reported differences in emissions per kg of DMI, although they found no differences in CH 4 production [30,33]. However, other studies reported lower CH 4 emissions and lower CH 4 emission per DMI in negative RFI animals [29,32]. One explanation would be a lower particle passage rate in the rumen due to differences in feeding behavior since more efficient animals spend less time feeding and therefore exhibit a higher feeding rate than positive RFI animals [19,32]. Residual feed intake is an intrinsic trait of the individual that reflects maintenance requirements [36]; thus, another explanation for the lower CH 4 emissions would be a lower energy intake of negative RFI animals compared to positive RFI animals [29], since CH 4 emission is positively associated with energy intake and differences in digestibility, CH 4 emissions, heat production and energy retention are the main factors responsible for the variation of RFI between animals [13].
Lower or equal CH 4 emission (g/day), but higher emission per kg of DMI (CH 4 /DMI) and a higher percentage of gross energy lost as CH 4 , in negative RFI animals compared to positive RFI were observed in the present study and have also been reported by other authors [30,33,34]. The reasons for these differences in CH 4 /DMI and CH 4 /GE between RFI classes  [19] found lower digestibility (4.7% to 9%) of dry matter, neutral detergent fiber and acid detergent fiber, as well as of crude protein.
The lower CH 4 emissions per kg of live and metabolic weight observed in animals classified as negative RFI (P = 0.0096 and P = 0.0033 for CH 4 /MBW and CH 4 /BW 0.75 , respectively) were similar to the results reported by Nkrumah et al. [13] and Fitzsimons et al. [28]. These findings indicate that, regardless of the effects of DMI, potential selection of cattle for RFI and reduced enteric CH 4 emission is possible [16,28]. The strong relationship between CH 4 and DMI and the divergent results reported in the literature for RFI and CH 4 emissions underscore the lack of evidence of a direct effect of RFI on enteric methane emission [35].
Considering the number of animals evaluated in this study and the fact that negative and positive RFI animals were compared (and not only extreme animals), it is possible to state that CH 4 emissions per kg of live body weight were different, with lower values in more efficient animals. This confirms the hypothesis that negative RFI animals emit less CH 4 per day or per kg of live body weight or per kg of ADG and, furthermore, these animals have strong potential to emit less CH 4 per kg of carcass. Considering that live body weight in the present study is the yearling weight, and the evidence of high genetic (0.55 to 0.89) and phenotypic (0.67 to 0.72) correlation between yearling weight and carcass weight [39-41], live body weight is a real indicator of the carcass weight.
Differences in the number of methanogenics in the rumen between more and less efficient animals may explain the differences observed in the intensity of CH 4 emission, regardless of the diet supplied [42]. In fact, Lopes et al. [43] and Andrade et al. [44] reported differences in the microbial composition of fecal samples between Nellore animals classified as negative and positive RFI. However, these differences in microbial populations between negative and positive RFI cattle may be due to differences in the ruminal passage rate and digestion as a result of different DMI levels [45]. Another approach to explain the variation in CH 4 emission between more and less efficient animals independent of DMI would be to identify differences in the efficiency of feed utilization, such as nutrient absorption, appetite regulation, and cell metabolism [35].
In conclusion, more efficient animals emit less CH 4 expressed as g/day and per kg of live body weight than less efficient animals, suggesting lower emission per kg of carcass. However, it is not possible to state that RFI has a direct effect on enteric CH 4 emissions since emission per kg of consumed dry matter and the percentage of gross energy lost as CH 4 were greater for negative RFI animals.
Supporting information S1