Feed efficiency in cattle

Residual feed intake (RFI) is a measure of feed efficiency. It is the difference between the amount of feed eaten by an animal over a test period and the amount of feed predicted or expected to have been eaten by the animal based on its size and level of production. By this measure of efficiency, animals that eat less than expected have a negative RFI and are considered more efficient. The trait is heritable in beef cattle and it is demonstrably possible to breed cattle for lower RFI. In Angus cattle, selection for RFI improved the efficiency of growth of young bulls and heifers, of steers at pasture and in the feedlot, and of mature non-pregnant, non-lactating cows fed a pelleted feed ad libitum. Differences in RFI are accompanied by differences in body composition; the genetic relationship between these variables appears stronger than the phenotypic relationship, but body composition alone is not the only driver of variation in RFI. The carcass of the high-efficiency steer is likely to have slightly lower subcutaneous fat depths at the rib and rump sites used for valuing the carcass, and compliance with increasingly strict specifications for fat finish will need to be monitored. Commercial beef producers can expect low-RFI cows to be heavier and leaner (slightly less subcutaneous fat). To date, no significant effect of RFI on maternal productivity traits has been observed, but there is a trend for high-efficiency cows to calve later in the season, which is likely associated with delayed onset of puberty. Other physiological changes are minor relative to the effects on fatness. There is increasing evidence that the variance in RFI is related to feeding level, which could affect the potential for genetic improvement of RFI under pasture conditions. Interactions between selection for lower RFI and feed conversion in cows grazing pastures that vary seasonally in availability and quality, cow body condition (fat) and cow lifetime productivity are still poorly understood and are the subjects of ongoing studies.

Residual feed intake (RFI) is a measure of feed efficiency. It is the difference between the amount of feed eaten by an animal over a test period and the amount of feed predicted or expected to have been eaten by the animal based on its size and level of production. By this measure of efficiency, animals that eat less than expected have a negative RFI and are considered more efficient.

RFI is heritable in beef cattle and it is demonstrably possible to breed low-RFI cattle i.e., cattle that eat less with no compromise in growth performance (Arthur and Herd, 2008). This reduction in feed intake would also have environmental benefits, viz., reduction in greenhouse-gas emissions, nitrogen excretion and manure production (Herd et al., 2002a; Arthur et al., 2010). It would appear that selection for lower RFI is an opportunity to simultaneously improve profitability and the environmental credentials of beef production, an opportunity that is attracting international attention.

In Australia, the major cattle breed societies have adopted RFI for the purpose of genetic improvement in feed efficiency. The term "net feed intake" refers to feed intake minus that expected based on an animal´s size and level of production (typically growth rate in cattle), and is equivalent to RFI. Net feed intake is used in place of RFI in this industry application. Estimated breeding values (EBVs) describe the genetic merit of an animal for breeding purposes. Trial EBVs for RFI were first published in 1999 for the Australian Angus breed, and for the Australian Hereford and Poll Hereford breed in 2002. Two EBVs for RFI have been published: an EBV for RFI of young animals (typically young bulls and heifers) tested post-weaning (EBVrfi-pw) and an EBV for RFI of animals (typically steers) that are fed a feedlot diet (EBVrfi-f). There is currently no RFI EBV for cows (BREEDPLAN, 2010).

Herein we review RFI in the context of its potential use for reducing feed costs. Evidence for variation in biological processes likely to contribute to differences in efficiency is discussed, with an emphasis on favourable, and potentially unfavourable, outcomes for farm animal productivity and profitability. We attempt to summarise the likely characteristics of an "improved" steer being fed for slaughter and those of an "improved" cow retained in the herd of a commercial cattle producer

Feed intake and utilization involve a complex of biological processes and pathways and interactions with the environment. That individuals of the same live weight require widely different amounts of feed to achieve the same level of production was acknowledged by Byerly (1941) in his experiments to determine whether individual differences in the net efficiency of laying hens are inherited. RFI was one of several indices used for calculating the feed efficiency of growing cattle by Koch et al. (1963) and recognized that differences in both body weight maintained and body weight gained affect feed requirements. They suggested that feed intake could be adjusted for body weight and body weight gain, effectively partitioning feed intake into two components: feed intake expected for the given level of production and a residual portion. The residual portion of feed intake was used to identify animals that deviated from their expected level of feed intake, and was moderately heritable; the more efficient animals had lower (negative) RFIs.

Since the work of Koch et al. (1963), individual variation in feed intake above and below that expected or predicted on the basis of size and growth rate have been demonstrated for mice (Archer et al., 1998), poultry (Luiting and Urff, 1991), pigs (Foster et al., 1983; Gilbert et al., 2007; Hoque et al., 2007) and beef cattle (Archer et al., 1999). There is also evidence for a genetic basis to this variation across these species; estimates for the heritability of RFI in the aforementioned reports and others range from low to moderate (see review by Pitchford, 2004). The only estimates of zero or nearzero heritability for RFI have been for high-producing dairy cows (Archer et al., 1999). In 2004, Pitchford suggested that these low estimates may reflect the method of feeding or the small size of the dataset, but as discussed in a subsequent section, they may be related to net energy intake.

Interest in breeding to improve beef cattle feed efficiency has increased greatly during the past decade (Herd et al., 2003a; Arthur and Herd, 2008; Herd, 2008). RFI may be less influenced by pre-test environmental factors than growth-related traits (Herd and Bishop, 2000; Arthur et al., 2001c). As this measure of feed efficiency is independent of an animal´s size and growth rate, variation in RFI may represent inherent variation in basic metabolic processes that determine production efficiency (Archer et al., 1999).

Direct and correlated responses to selection for RFI were studied using divergent selection lines of Angus cattle established at the Trangie Agricultural Research Centre, NSW, Australia. Creation of the RFI selection lines commenced in 1994. The 1993- and 1994-born cattle formed the foundation herd. After completing a post-weaning RFI (RFIpw) test, the females were allocated to selection lines based on individual RFI values. Those with low RFI (< 0; more efficient) were allocated to the low-RFI line and those with high RFI (>0; less efficient) were allocated to the high-RFI line. Three to six of the lowest and highest RFI bulls tested each year were allocated to the low- and high-RFI lines, respectively. Divergent selection, based solely on individual RFIpw, continued until 1998; the first progeny of the selected parents were born in 1995 and the last were born in 1999. During the RFI tests, animals had ad libitum access to a pelleted ration consisting of 70% lucerne hay and 30% wheat containing 10.5 MJ metabolisable energy (ME) per kg dry matter (DM) and 16% crude protein. The RFI test procedure and the design of the selection lines are described in detail by Arthur et al. (2001a, b).

Direct and correlated responses in post-weaning feed efficiency and growth traits after 5 years of divergent selection for RFIpw (approximately two generations) are summarised in Table 1. Compared with the high-RFI line progeny, the low-RFI line progeny had significantly lower RFI (−0.54 vs 0.71 kg/d, respectively), lower feed intake (9.4 vs 10.6 kg/d, respectively) and better feed conversion (6.6 vs 7.8, respectively), but did not differ in average daily gain (ADG; 1.44 and 1.40 kg/d, respectively) over the test period or in final live weight (384 and 381 kg, respectively). These results show that selection for RFIpw improved post-weaning feed efficiency with minimal effect on growth. The 1.247 kg/d divergence between lines in feed consumed represented a 14% saving in feed cost (Arthur et al., 2001a).

Residual feed intake measures whether an animal eats more or less feed than predicted according to published feeding standards or feed intake records. In the late 19th century, it was established that farm animals follow the physical laws of conservation of mass and energy. It follows that, apart from measurement error in its component traits (e.g., feed intake, live weight, weight gain), variation in RFI must be underpinned by measurable differences in biological processes.

Studies on lines of animals selected for other traits (e.g., growth rate, wool production) indicate that no single mechanism is likely to be primarily responsible for the associated change in phenotype (Oddy, 1999). For example, replication of lines of mice selected for divergence in growth rate resulted in mice with similar divergence in the selected trait, but markedly different phenotypes with respect to body composition, feed intake, metabolism and activity (Falconer, 1973). There are a limited number of cases in which a single gene mutation has caused a marked phenotypic difference in livestock species, for example, the mutation in the myostatin gene that causes the double-muscled phenotype in cattle (Grobet et al., 1997). However, is apparent that many physiological mechanisms contribute to variation in RFI. This was shown in experiments on Angus steer progeny after a single generation of divergent selection for RFIpw (Richardson and Herd, 2004). The difference in energy retained as protein and fat accounted for only 5% of the difference in RFI following divergent selection. Of the variation in RFI, digestion accounted for 10%, feeding pattern for 2%, heat increment of feeding for 9% and activity for 10%. Indirect measures of protein turnover suggest that protein turnover, tissue metabolism and the stress response accounted for at least 37% of the variation in RFI. Approximately 27% of the difference in RFI was due to variation in other processes that were not measured, e.g., ion transport (Figure 1). A comprehensive review of the physiological basis of variation in RFI, including recent genomic evidence, was published by Herd and Arthur (2009).

Table 1. Trangie residual feed intake (RFI) divergent selection lines: means for production traits for young bulls and heifers post-weaning and for steers after divergent selection for low RFI measured post-weaning (RFIpw; high efficiency) or high RFIpw (low efficiency), and regression coefficient with their estimated breeding value for RFIpw (EBVrfi-pw).

In this review, we have used the Trangie low-RFI selection line as a phenotypic model of high feed efficiency and the high-RFI selection line as a phenotypic model of low feed efficiency. These closed lines of cattle were the progeny of parents selected on the basis of post-weaning RFI during ad libitum consumption of a medium quality (10 MJ ME/kg DM) pelleted diet (Arthur et al. 2001a).

Figure 1. Contributions of biological mechanisms to variation in residual feed intake as determined from experiments on divergently selected young cattle (from Richardson and Herd, 2004).

High-efficiency steers grew a little faster than low-efficiency steers but there was no difference in pasture intake during grow-out on summer pasture (Table 1; Herd et al., 2002b). Feed conversion ratio (FCR) appeared to be two units lower (better) in the low- RFI selection line steers but this difference was only significant at P = 0.1. These associations were consistent with the regression coefficients for these traits against the mid-parent EBVrfi-pw of the progeny, i.e., negative for ADG (progeny of parents with a negative mean EBVrfi-pw grew faster), not statistically significant for pasture intake and positive for FCR (progeny of parents with a negative mean EBVrfi-pw had lower, i.e., better, FCR). Comparison of alkane profiles in pasture and faeces indicated that the steers mainly consumed perennial ryegrass and some tall fescue, with a trend (P < 0.1) for the steer progeny of high-efficiency parents to have a higher proportion of ryegrass in their diet. There was no evidence that digestibility differed between the selection-line progeny groups or that it varied with RFI.

Further evidence for a favourable association of steer growth and feed efficiency at pasture with genetic variation in RFI was reported by Herd et al. (2004), who studied the growth and feed intake of Angus and Hereford weaner steer progeny of sires with known EBVs for RFI-f on three different pasture systems during spring and summer in northern New South Wales. Significant regression coefficients with sire EBVrfi provided evidence for favourable associations between ADG, pasture RFI and FCR. The lack of a significant regression coefficient between pasture intake and sire EBVrfi indicated that superior FCR was caused by superior ADG, and not lower feed intake. The results showed that a 1 kg/d lower sire EBVrfi was associated with steer progeny that grew 19% faster, with no increase in pasture consumption, a 26% lower RFI and a 41% better FCR. Trangie high-efficiency steers are often a little heavier and leaner (lower subcutaneous fat depths over the rib and rump scanning sites) at the end of a backgrounding phase at pasture prior to feedlot entry.

In both experiments, growth of the steers at pasture was well below their genetic potential. The lack of a difference in pasture intake provided evidence that other factors, such as pasture characteristics, were regulating feed intake. Superior FCR resulted from superior ADG when the pasture conditions limited the growth rates of the steers. There are no studies on the foraging behaviour at pasture of high-efficiency cattle. Differences in activity have been observed between high- and low-efficiency steers confined to feedlot yards (Richardson et al., 1999), as have differences in patterns of feeding from automated feed dispensers/ recorders when feed consumption is not restricted (Richardson, 2003; Robinson and Oddy, 2004; Dobos and Herd, 2008). The energy cost of work involved in feeding, ruminating and locomotion has been estimated for high- and low-RFI selection-line bulls and heifers under standard test conditions, and accounted for approximately 5% of the increase in feed energy intake of the high-RFI cattle (Herd et al., 2004). Whether there are differences in activity or foraging behaviour at pasture between high and low efficiency cattle remains to be determined.

Studies on monogastric species have demonstrated the contribution of differences in activity to variation in RFI. In pigs, total daily feeding time and the number of visits to a feeding station are positively correlated with RFI (r = 0.64 and 0.51, respectively; de Haer et al., 1993). Activity contributes to a substantial proportion of the variation in RFI in chickens (Braastad and Katle, 1989; Katle, 1991; Luiting et al., 1991). Luiting et al. (1991) concluded that 80% of the genetic difference in RFI between lines of chickens divergent for RFI could be related to a difference in physical activity. In lines of mice divergently selected for heat loss, Mousel et al. (2001) showed that the high heat loss (low efficiency) mice were twice as active as the low heat loss (high efficiency) mice and that this difference in activity accounted for 36% of the difference in feed intake between the selection lines. In lines of mice divergently selected for food intake corrected for body weight, Bünger et al. (1998) found that mice in the high food intake (low efficiency) line were three times more active than mice in the low feed intake (high efficiency) line.

Feedlot performance and carcase attributes of steer progeny from parents selected for low or high RFIpw were investigated using three cohorts of Trangie-bred calves born in 1997, 1998 and 1999 and fed for slaughter at light, heavy and medium market weights, respectively (Herd et al., 2003b). Selection for low RFIpw produced steer progeny that ate less per unit live weight gain compared with steers from high RFIpw parents over a 70-day test period in the feedlot, with no adverse effects on growth (Table 1). Low-RFI line steers tended to have lower (better) FCR than the high-RFI line steers (7.6 vs 8.2, respectively) and had lower RFI (–0.12 vs 0.10 kg/d). Significant positive regressions of FCR and RFI with EBVrfi-pw provided further evidence for favourable associations with genetic variation in postweaning RFI.

Growth, feed intake over 251 days, feed efficiency and carcass traits of 208 yearling Angus steers differing in genetic merit for RFI were measured in a large commercial feedlot (Herd et al., 2009). At feedlot entry, the steers were drafted into one of three groups: high efficiency (HE; midparent EBVrfi-pw less than or equal to –0.3 kg/d), medium efficiency (ME: midparent EBVrfi-pw –0.3 to 0.14 kg/d) and low efficiency (LE; midparent EBVrfi-pw >0.14 kg/d). The HE steers grew as fast as or faster than the ME and LE steers (1.11 kg/d, 1.06 kg/d and 1.07 kg/d, respectively). At slaughter, the HE and ME groups were heavier than the LE group (713 kg, 714 kg and 701 kg, respectively). The HE steers consumed less feed than the ME and LE steers over the 251 days (10.4 kg/d, 11.8 kg/d and 11.1kg/d, respectively). Compared with the LE steers, the HE steers had a 10% lower FCR and a 0.98 kg/d lower RFI. The experiment showed that genetic superiority in RFI reduced feed consumed over 251 days in a large commercial feedlot with no compromise in weight gain.

Genetic variation in total tract digestion of feed contributes to between-animal differences in feed intake. For example, in ewes from lines of sheep selected for or against weaning weight, the magnitude of the difference was about two percentage units of organic matter digestibility around a mean of 70% (Herd et al., 1993). In these selection lines, digestibility in 16-month-old rams was four percentage units higher in the high weaning weight line than in the low weaning weight line at a level of feeding close to ad libitum (Oddy, 1993). Wilkes et al. (2011) compared Damara and Merino sheep and found that the Damara had a much higher digestibility (52%) than the Merino (42%) when fed low-quality feed. Whereas the feed intake of the Damara was only 2% greater than that of the Merino, its RFI was 7% lower (better) than that of the Merino. This illustrates the contribution of variation in digestibility to variation in RFI.

Richardson et al. (1996) found that young bulls and heifers that phenotypically ranked low or high for RFIpw tended to differ in their ability to digest DM by about one percentage unit when tested using a pelleted diet with an estimated DM digestibility of 68%. This difference in DM digestibility accounted for 14% of the difference in intake between the two groups of cattle. Digestibility was correlated with RFI in cattle fed a high-grain diet in individual pens in an animal house. The magnitude of the correlation (r = –0.44) indicates that differences in digestibility accounted for 19% of the phenotypic variation in RFI and the direction of the correlation indicates that lower RFI (higher efficiency) was associated with higher digestibility (Richardson and Herd, 2004). The difficulty in precisely measuring small differences in digestibility suggests that caution should be exercised in assuming that variation in digestion is a major factor contributing to differences in RFI in beef cattle. Studies on monogastrics indicate that differences in digestibility are not important sources of variation in RFI in chickens (Luiting et al., 1994), pigs (de Haer et al., 1993) or mice (Bünger et al., 1998).

The rate of feed ingestion and the duration of the meal have been described as key factors determining the energy cost of eating in cattle (Adam et al., 1984). A study of feeding patterns of Angus steers bred for high or low RFI (Richardson, 2003) reported a trend (P < 0.10) for high-RFI steers to have a faster decline in average daily feeding session times over their feed intake test and to spend more time eating early in the test compared with low-RFI steers. Spectral analysis of feeding patterns for another cohort of Angus steers from the RFI selection lines found that the high-RFI steers had more variable temporal patterns of feed intake early in the RFI test period compared with low-RFI steers, with the latter appearing to quickly settle into a regular feed intake cycle (Dobos and Herd, 2008). Robinson and Oddy (2004) reported genetic variation in three feeding behaviour traits of feedlot steers, that they had moderate heritabilities and were positively correlated with RFI. The higher feed intake accompanying higher RFI was associated with a longer feeding time per day, more eating sessions per day and a faster rate of eating (g/min). Feeding time and number of eating sessions (but not eating rate) also had positive genetic correlations with RFI, indicating that effects of some genes for these feeding behaviours also had an effect on effect on RFI. Montanholi et al. (2010) also found that high-RFI (less efficient) steers had a larger meal size, a higher eating rate and made more visits to the feeder than low-RFI steers. Less efficient steers (high-RFI) made more visits to the feeder during the night than more efficient steers. Differences in activity between high- and low-efficiency steers in feedlot yards was examined by Richardson et al. (1999). They reported a phenotypic correlation of 0.32 between RFI and "daily pedometer count", which indicates that about 10% of the observed variation in RFI was explained by this measure of activity.

Richardson and Herd (2004) hypothesized that susceptibility to stress is a key driver for many of the biological differences observed in beef cattle following divergent selection for RFI. This hypothesis is supported by several measurements that indicate that high-RFI (low efficiency) steers are more susceptible to stress than low-RFI (high efficiency) steers, and as a consequence, metabolize more feed energy than predicted on the basis of weight or weight gain. Montanholi et al. (2010) reported that low-RFI (more efficient) steers had lower cheek and snout temperatures than less efficient steers (28.1 °C vs 29.2 °C, respectively, and 30.0 °C vs 31.2 °C, respectively), indicating greater energetic efficiency for the low-RFI steers. Plasma cortisol level was not correlated with efficiency traits but more efficient steers had higher fecal cortisol metabolite levels than less efficient steers (51.1 vs 31.2 ng/g, respectively), indicating that a higher cortisol baseline is related to better feed efficiency.

In laying hens from lines subjected to divergent selection for RFI, the more efficient hens had a significantly lower corticosterone maximum response to an exogenous adrenocorticotrophic hormone (ACTH) challenge than the low efficiency line hens, but the response of the more efficient hens was sustained for longer than that of the less efficient hens (Luiting et al., 1994). It has also been suggested that many of the differences in the activity of chickens divergently selected for RFI reflect differences in frustration behaviour associated with long-term stress (Luiting et al., 1994). Altan et al. (2004) found that low-RFI quail were less fearful and less susceptible to stress. In young crossbred rams, poor feed efficiency (measured as either high RFI or high FCR) was phenotypically associated with higher basal levels of serum cortisol and a greater elevation in serum cortisol level during an ACTH challenge (Knott et al., 2008). Genetic variation in susceptibility to stress has been reported in pigs (Zhuchaev et al., 1996). High rates of cortisol production in pigs have been associated with frustration at the lack of control over or predictability of their environment (Dantzer, 1981). Given that the physiological responses to stress include an increase in metabolic rate and energy consumption coupled with an increase in catabolic processes such as lipolysis and protein degradation (Knott et al., 2008), variation in stress response warrants further evaluation as a potential mechanism involved in differences in feed efficiency.

Evidence exists that there is a genetic relationship between RFI and subcutaneous fat depth, i.e., more efficient (lower RFI) animals are leaner than less efficient (high RFI) animals. Ultrasound measurement before slaughter showed that low-RFI line steers had less fat over the standard Australian 12/13th rib and P8 rump sites and a smaller eye-muscle area than high-RFI line steers (Herd et al., 2003b). The low-RFI steers had a similar carcass weight to the high-RFI steers, less fat depth at the rump on the hot carcass (14.9 vs 16.5 mm, respectively; P < 0.05) and a slightly lower dressing percentage. Significant (P < 0.05) regressions for the subcutaneous fat measurements, eye-muscle area and dressing percentage with EBVrfi-pw provides additional evidence for the association of these carcass traits with genetic variation in feed efficiency (Herd et al., 2003b). No compromise in intramuscular fat content (marbling) has been reported (McDonagh et al., 2001). Three groups of Angus steers differing in midparent EBVrfipw (high efficiency, HE; medium efficiency, ME; and low efficiency, LE) were slaughtered after they were fed for 251 days in a large commercial feedlot (Herd et al., 2009). Carcass weight and dressing percentage were lowest in the LE steers and eye muscle area was highest in the ME steers. Fat depth over the ribs of the carcass was lower in the HE steers than in the ME and LE steers (15.6 mm, 17.6 mm and 20.7 mm, respectively). AUSMEAT and USDA marbling scores (AUSMEAT: 1 to 9 by 0.1unit scale, Anon, 2005; USDA: 100 to 1100 by units of 10, Romans et al., 1994) were highest in the ME group and did not differ between the HE and LE groups, demonstrating that correlations with fatness are not the same for all fat depots (Egarr et al., 2009). The experiment showed that genetic superiority in RFI was accompanied by lower (better) FCR over 251 days of feeding in a large commercial feedlot with no compromise in carcass weight, dressing percentage or marbling grade.

Total tissue dissections of the carcasses of steers subjected to divergent selection for RFIpw (Richardson et al., 2001) showed significant selection-line differences in dissectible carcass fat and total dissectible fat as a percentage of bodyweight (Table 1), which is in agreement with the genetic correlation between RFIpw and subcutaneous fat depth. There was a significant positive regression coefficient between retail beef yield and EBVrfi-pw, which is favourable providing market specifications for fatness are met.

In the few instances in which contributions of body composition to genetic variation in heat production or feed efficiency have been studied, it was found that variation in body composition was small relative to variation in heat production (Herd et al., 2004). Results for beef steers divergently selected for RFI (Richardson et al., 2001) showed that whole-body chemical composition was correlated with genetic variation in RFI, viz., steer progeny of low-RFI parents had less whole-body fat and more whole-body protein than progeny of high-RFI parents. The differences in energy retained in the body were estimated to account for only 5% of the difference in feed intake, with the remainder (95%) due to heat production.

The size and direction of genetic correlations between measures of body composition and RFI provide additional information on the magnitude of the association between body composition and variation in RFI. Arthur et al. (2001b) found that subcutaneous fat depth over the 12/13th rib and the P8 rump site had positive genetic correlations with RFI of 0.17 and 0.06, respectively, in beef weaner bulls and heifers. For yearling bulls of several beef breeds, Schenkel et al. (2004) reported genetic correlations between RFI and fatness traits of similar magnitude to those reported by Arthur et al. (2001b), viz., r = 0.16 for scanned backfat thickness and r = –0.02 for scanned intramuscular fat percentage (IMF%). In these young cattle, although these measures of body fat had statistically significant correlations with genetic variation in RFI, they explained less than 5% of the variation in RFI. In young feedlot steers, Nkrumah et al. (2007) reported slightly stronger genetic correlations for phenotypic RFI with backfat thickness and marbling fat score, both for measurements on the live animal (r = 0.35 and r = 0.32, respectively) and measurements on the carcass (r = 0.33 and 0.28, respectively). In older feedlot steers and heifers, Robinson and Oddy (2004) reported genetic correlations between the 12/13th rib and rump fat depths and RFI of of 0.48 and 0.72, respectively, and a genetic correlation of 0.22 between IMF% and RFI; evidence for a much stronger association between the effect of genes controlling these measures of fatness and their effect on RFI. In pigs, where attainment of moderate levels of fatness in the carcass is also required, a moderately strong and antagonistic genetic relationship between RFI and carcass backfat thickness (r = 0.44) was reported by Gilbert et al. (2007).

In chickens, reports vary as to the contribution of differences in body composition to variation in RFI. Luiting (1990) summarized reported genetic and phenotypic correlations of body fat traits with RFI, which ranged from –0.40 to 0.45. In a later paper, Luiting et al. (1991) found that the low-RFI line contained 3.4% more fat than the high-RFI line. Subsequently, Gabarrou and Picard (1997) and Tixier-Boichard et al. (2002) both reported lower body fatness in the high RFI lines, that is, the opposite to that observed in cattle and pigs. In mice, improved RFI was associated with a slight increase in fat post-weaning and a decrease in fat at maturity (Archer et al., 1998).

These results suggest that the magnitude of the association between body composition and variation in RFI is influenced by the age and stage of maturity of the test animals. It may be that performance tests on beef cattle and pigs usually involve growing animals, in which protein synthesis is more efficient than fat deposition, whereas, in the adult animal, maintenance requirements for protein are higher are than for fat, favouring an association between increased fatness and lower RFI, which is more typical of results with poultry and mice (Tixier-Boichard et al. 2002).

After a single generation of divergent selection for post-weaning RFI, meat samples taken from the M. longissimus dorsi (LD) of feedlot-finished steers showed no differences between selection lines in shear force or compression after 1 or 14 days of aging, or in initial muscle concentrations of m- and μ-calpain (enzymes associated with initiation of muscle fibre breakdown) (McDonagh et al. 2001). However, LD muscle from low-RFI steers contained a slightly higher concentration of calpastatin (an inhibitor of calpain) and a lower level of myofibre fragmentation than LD muscle from high- RFI steers. These results provide evidence that ongoing selection for low RFI (high efficiency) could negatively affect meat tenderness. This association should be monitored. Small differences in myofibre fragmentation are consistent with the proposal that differences in protein degradation and turnover contribute to variation in RFI (Richardson and Herd, 2004). Recent unpublished results from later generations of progeny in the feedlot trial referred to above (Egarr et al., 2009; Herd et al., 2009) show that meat from low-RFI steers had a greater peak force and a greater aging rate than that from high-RFI steers. Consequently, there was no difference between these lines in tenderness after aging for 7 days. The low-RFI steers had less calpastatin activity and therefore had a greater rate of proteolysis than high-RFI steers, suggesting that protein turnover is not the cause of differences in RFI.

The Trangie selection trial indicates that commercial beef producers who for low RFI can expect an animal that has superior growth rate on reasonable quality pastures, differs little from unselected animals in the amount of pasture consumed, and has superior FCR. The high-efficiency steer is likely to be a little heavier and leaner (lower subcutaneous fat) at the end of a backgrounding phase at pasture prior to feedlot entry. Foraging behaviour, especially on lower quality (low availability, low digestibility) pastures, and associated differences in animal performance remain to be studied. In the feedlot, the high-efficiency steer should grow faster and eat less with no compromise in final weight, which would reduce feed costs. The carcass of the high-efficiency steer is likely to have slightly less subcutaneous fat depth at the rib and rump sites used for valuing the carcass, and compliance with increasingly strict specifications for fat finish should be monitored. The high-efficiency steer carcass may have a slightly greater yield of retail beef but may not have an advantage in dressing percentage. No compromise in intramuscular fat content (marbling) has been reported. According to objective measurements made using an instrument, meat tenderness is reduced in high-efficiency steers, but the Angus steers studied all produced meat that would have been regarded as tender by Australian consumers.

Research to improve feed efficiency has mainly been directed at growing cattle (typically steers) consuming high-cost grain-based diets. In a commercial beef cattle herd, the feed energy consumed by the cows constitutes 75% or more of the total feed consumed annually by the herd, and maintenance represents 60–75% of the total energy requirements of individual breeding cows (Archer et al., 1999). The impact of RFI on the cost of maintaining cows is clearly important for the efficiency and profitability of beef production systems.

Cows from the Trangie RFI selection lines provide a phenotypic model of the likely impact of selection for RFI on the female component of a beef herd. After the post-weaning test, all heifers entered the cow herd. Cows were not mated again after the birth of their second calf and were re-tested for RFIcow in a manner similar to the post-weaning RFI test about 10 weeks after the calf was weaned. One cohort of females was tested for efficiency as lactating 3-year-old cows at pasture and as 4-year-old non-pregnant dry cows at a restricted level of feeding before being tested for RFI as cows. After the RFIcow test, cows entered the Trangie herd and their reproductive performance was recorded over subsequent years.

The mature cow test traits were all moderately to highly heritable (Archer et al., 2002). The results indicated that there was significant genetic variation in daily feed intake and feed efficiency (RFI and FCR) between the cows. The phenotypic and genetic relationships between traits measured during the postweaning and mature cow tests are presented in Table 2. At the phenotypic level, most traits (with the exception of metabolic mid-test live weight) were only moderately correlated from post-weaning heifers to mature cows. However, at the genetic level, all traits, with the exception of FCR, were highly correlated across the two ages. In particular, the relationships between postweaning and mature daily feed intake and RFI were strong, and the genetic correlations approached unity. Therefore, selection for lower post-weaning RFI will reduce the intake of a pelleted ration by dry, nonpregnant cows and slightly increase cow weight, thus improving the efficiency of the cow herd.

These strong relationships present an opportunity to utilise selection for improved feed efficiency in young growing animals to simultaneously improve feed efficiency in mature cows fed at an ad libitum level. However, Archer et al. (1999) stated that information is required on whether animals selected under conditions typical of those used in RFI tests are also superior when offered diets typical of those consumed by cows at pasture. To address this issue, Herd et al. (2011) conducted an experiment on the phenotypic associations between efficiency traits in a cohort of Angus females that were tested as heifers for RFIpw, as lactating 3-year-old cows at pasture, at a restricted level of feeding as 4-year-old non-pregnant dry cows and were then tested for RFIcow at an ad libitum level of feeding. Phenotypically, lower RFI during the RFIpw test was associated with heavier cow weight during lactation 2 years later, with no associated increase in pasture intake, and with a trend to lower (better) feed efficiency at pasture (Table 2). However, post-weaning RFI was not associated with variation in efficiency (as ADG or RFI) during the restricted feeding test. Lower (better) RFIpw was associated with lower RFI, but not FCR, in 4-year-old dry cows fed ad libitum. Superior efficiency at pasture was associated with inferior efficiency as indicated by lower ADG and was not associated with RFI in the subsequent restricted feeding test or with efficiency in the mature cow RFI test. Efficiency in the restricted feeding test was not associated with efficiency, as RFI or FCR, in the mature cow RFI test. RFI at a restricted level of feeding is discussed in a subsequent section. In summary, there was evidence that heifers identified as phenotypically superior in feed efficiency under ad libitum conditions post-weaning were also superior as lactating cows when grazing medium-quality pasture and as dry cows under ad libitum conditions, but not when tested for efficiency at a level just above maintenance. Superior efficiency at a restricted level of feeding was not phenotypically associated with superior efficiency in the other three efficiency tests. At the genetic level, lower (better) post-weaning RFI was associated with lower (better) phenotypic RFI during post-weaning RFI testing, with heavier lactating cow weight at pasture but not with superior feed efficiency during restricted feeding, and was associated with lower (better) FCR and RFI in the mature cow RFI test.

Table 2. Correlation coefficients (r-values) between growth, feed intake and efficiency traits for Angus heifers in post-weaning residual feed intake (RFI) tests and as cows in a pasture efficiency test, a restricted feeding efficiency test and mature cow RFI tests.

In a review of results from several species, Pitchford (2004) concluded that low RFI might have negative effects on reproduction. Since that time, two studies have been conducted on the maternal productivity of cows from the Trangie RFI divergent selection lines. In the first, data from 185 Angus cows were used to study the effect of divergent selection for RFI on maternal productivity across three mating seasons, starting in 2000 (Arthur et al., 2005). The cows were the result of 1 to 2.5 generations of selection (mean = 1.5 generations), and differed in RFIpw EBV by 0.8 kg/d. The average RFIpw EBV of cows in the low- RFI line differed significantly from that of the high- RFI cows (–0.254 ± 0.287 (sd) kg/d and 0.533 ± 0.314 kg/d respectively; P < 0.01; P.F. Arthur, pers. comm.). In general, the cows lost subcutaneous fat (measured twice per year) when they were nursing calves and gained fat thereafter. High-RFI cows had significantly (P < 0.05) greater rib fat depths than low-RFI cows at the start of the 2000 (10.8 mm vs 9.3 mm, respectively), 2001 (11.3 mm vs 9.8 mm, respectively) and 2002 (7.0 mm vs 5.7 mm, respectively) mating seasons. No other significant selection-line differences in fatness were observed. No significant selection line differences in weight (measured four times per year) were observed. However, the cows either maintained or lost weight during lactation and gained weight thereafter; mean weight ranged from 450 to 658 kg. There were no significant selection-line differences in pregnancy rate (90.4%), calving rate (88.7%), weaning rate (80.8%) or milk yield (7.7 kg/d; Table 3). There was a trend (P < 0.1) for high-efficiency cows to calve slightly later than low-efficiency cows, but this was not associated with a difference in the weight of calf weaned per cow exposed to the bull (mean = 195 kg).

The second study on the maternal productivity of cows was conducted to evaluate early-life reproductive performance and the onset of puberty in females from the Trangie RFI-divergent selection lines (Donoghue et al., 2011). In the first part of this study, data on 1999-born females (n = 64) were evaluated in terms of weight, subcutaneous fat (P8 fat depth) and reproductive performance over two breeding cycles. These females were the result of approximately 1.8 generations of selection and the mean EBVrfi-pw of their parents differed by 1.4 kg/d. As observed in the previous study, no significant selection-line differences were evident for cow weight, pregnancy rate, calving rate, calf birth weight or weight of calf born per female exposed to the bull (Table 3). Females from the low-RFI line had significantly (P < 0.05) lower P8 fat depth relative to their high-RFI contemporaries at most of the measurement dates. Low-RFI females calved significantly (P < 0.05) later in the calving season than high-RFI females (35.7 d vs 27.6 d, respectively). Basarab et al. (2007) also reported that low RFI cows calved 5–6 days later. These results indicate that there is a delayed pregnancy date during the first mating season, which resulted in a later calving date for the low-RFI heifers. The later first calving date was then maintained at the subsequent calving. The later calving did not affect pregnancy rate or calving rate. However, in commercial herds in which restricted mating is practiced, 8 days could represent an 8% difference in pregnancy rate for a breeding season (D.J. Johnston, pers. comm.). In the second part of this study, ultrasonography was used to scan the ovaries of 2008-born heifers on four occasions after weaning. In these heifers, the presence of a corpus luteum provided evidence of ovulation, and hence the onset of puberty. The mean EBVrfi-pw of their parents differed by 1.1 kg/d. Irrespective of selection line, heifers that had attained onset of puberty had significantly (P < 0.05) greater P8 fat depth than those that had not attained puberty (Donoghue et al., 2011). Hence, the expectation was that, relative to high-RFI heifers, the low-RFI heifers, which had less rump (P8) subcutaneous fat, would attain onset of puberty at a slightly older age. This expected trend was observed but the difference was not significant, and further investigations were recommended.

Table 3. Trangie residual feed intake (RFI) divergent selection lines: means for maternal productivity traits for cows from lines divergently selected for low RFI (high efficiency) or high RFI (low efficiency).

Strong relationships exist that present an opportunity to utilise selection to improve feed efficiency of growing animals and mature cows simultaneously, based on measurements made post-weaning. The Trangie selection trial indicates that commercial beef producers can expect high-efficiency cows to be heavier and slightly leaner (slightly less subcutaneous fat) than unselected cows. After approximately two generations of divergent selection for RFI, no significant selectionline differences have been observed for the key maternal productivity traits of pregnancy rate, calving rate, calf birth weight and weight of calf weaned per female mated. However, there is a trend for high-efficiency cows to calve later in the season than low-efficiency cows, which may be associated with a delay in the onset of puberty because they are slightly leaner at first joining. It is not known whether ongoing selection would be associated with a decline in pregnancy rate during a restricted mating period such as that used in commercial herds. Furthermore, the effectiveness of selection for lower RFI at a restricted level of nutrition such as that typical of pasture-based production systems for much of the year requires further research.

Previous studies have focussed on the heritability of RFI, selection response and correlated responses. However, variance in RFI has received little attention. The variance in feed intake is a function of how much variance there is in growth (live weight and daily gain) and how much of the variance in intake is associated with that variance in growth. We hypothesize that variance will be elevated under ad libitum feeding conditions and low at a restricted level of feeding. When there is negligible variance in RFI, there is negligible variation in efficiency and consequently, little opportunity for improvement.

Bordas et al. (1995) reported that differences between poultry RFI selection lines were greater on the regular control diet than on control diet diluted with 19% bran (and hence with a lower feed energy density). Selection for efficiency in fish is done by selecting for growth at a restricted level of feeding. Silverstein (2006) found that although genetic differences between families were present under satiation feeding conditions, they were absent under limited feeding conditions. Veerkamp et al. (1995) reported that variance in RFI in dairy cattle was much lower in early lactation (effectively, restricted feeding) than later in lactation.

Roberts et al. (2007) measured RFI in composite beef heifers and found that the variance was 22-fold greater under ad libitum feeding conditions (0.088 kg2/ d2) than under restricted feeding conditions (0.004 kg2/ d2). The restricted diet was set at 80% of that eaten by the ad libitum group. If the relationship between feeding level and variance in RFI were linear, then this would suggest zero variance in RFI at 75% of ad libitum. Lines et al. (2009) reported differences in energy metabolism between 16 heifers from RFI selection lines at two feeding levels. Subsequent analysis (unpublished) showed that the variance at ad libitum intake (6.9 kg/d) was 0.0610 kg2/d2 compared with 0.0017 kg2/d2 at just above maintenance (4.2 kg/d). This represents a 36-fold difference and is similar to that reported by Roberts et al. (2007). Variance in RFI of Angus cows (n = 56) was 0.66 kg2/d2 when fed ad libitum and 1.01 kg2/d2 when tested as mature cows (Herd et al., 2011). When fed at a restricted level of feeding just above maintenance, the variance in RFI was 0.27 kg2/d2.

Another recent trial at the University of Adelaide demonstrated a similar effect when intake was restricted by quality rather than quantity. Wilkes et al. (unpubl. data) the compared growth and energy balance of Merino and Damara lambs offered ad libitum access to a low or a high quality diet (7 or 11 MJ ME/kg DM, respectively). The variance in RFI was 0.38 kg2/d2 for the low quality diet and 1.67 kg2/d2 for the high quality diet. This 4.4- fold difference demonstrates the relationship between intake and variance in RFI regardless of the method of intake restriction. This is similar to the results of the study of Veerkamp et al. (1995), in which the variance in RFI of dairy cows was low in early lactation even though they were fed ad libitum throughout lactation.

The implication of these results is that variance in RFI is related to variance in net energy available to the animal. This supports findings that RFI is related to body composition and may warrant inclusion of a measure of body composition in the RFI equation. Unfortunately, this does not indicate that our current measure of RFI, based on live weight and ADG, reflects differences in metabolic efficiency, which was the original aim of selecting for the trait. If these results extend to adult ruminants exposed to annual variation in energy availability, then we may conclude that during times of energy restriction (either due to feed quality or quantity), there may be negligible variation in RFI. During times of high energy availability, there will be large variation in RFI, which is correlated with fatness and may be of little financial value because at these times pasture utilisation and therefore feed costs are often low.

To exploit genetic variation in feed efficiency requires incorporation of feed intake as an additional selection criterion into beef cattle breeding programs. However, collection of information on feed intake and growth performance under standardised conditions is expensive. It is therefore important that the investment in collecting RFI data is justified by the economic returns generated. As it is unlikely that routine measurement of the feed intakes of all candidate animals for breeding could be justified, Archer et al. (2004) evaluated breeding program designs that target investment to generate maximum returns.

Assuming that the relative cost of testing, feed and prices received for cattle have remained stable since their evaluation, Archer et al. (2004) showed that it would be profitable for the Australian beef cattle industry to incorporate measurement of (residual) feed intake on a proportion of candidate sires for the seedstock sector. These results were for two different beef production systems in southern Australia that targeted either the domestic market (pasture-based finishing) or the Japanese market (animals finished on grain). The analysis considered a breeding program incorporating most information sources currently used in beef cattle genetic evaluation in Australia and compared this with a program that included measurement of feed intake in a two-stage selection process. After accounting for the cost of measuring feed intake (which then ranged from $150 to $450), additional profit was generated from inclusion of feed intake measurement on a proportion of bulls for all breeding schemes considered. Profit was generally maximised where 10% to 20% of bulls were selected at weaning for measurement of feed intake, with the selection of the bulls being done on a multipletrait index incorporating all information available on the bulls and their relatives.

Selection for RFI is effective in improving the efficiency of growth of young bulls and heifers, of steers at pasture and in the feedlot, and of mature nonpregnant, non-lactating cows fed ad libitum on pelleted feed. Differences between selection lines in RFI are accompanied by differences in body composition but the phenotypic correlation is not strong.

The small improvement in the lean beef yield of steers afforded by selection for low RFI should be balanced against increasingly tight specifications for fat on the carcass. There is a need to ensure that selection for improved feed efficiency does not delay the onset of puberty in the young cow and that cow productivity is not compromised in commercial herds with a restricted joining season. The interactions between selection for lower RFI and improvement in feed conversion of cows grazing pastures that vary in availability and quality, cow body condition (fat) and cow lifetime productivity are poorly understood. This is the subject of the Beef CRC Maternal Efficiency Project.

Rauw (1998) proposed that genetic selection has increased the occurrence of behavioural, physiological and immunological problems in livestock. Estimation of RFI involves estimation of feed intake based on live weight and product output. The composition of the output is rarely considered, nor are differences in locomotion, disease status, immunocompetence or other metabolic processes that use energy. It follows that if no allowance is made for the energy requirements of these processes, the reduction in feed intake sought by selection for low RFI may compromise an animal´ s capacity to sustain these functions (Rauw, 1998).

There is a need for a better understanding of energy requirements for maintenance and production (including reproduction) and of variation in the efficiency of energy utilization for these processes throughout an animal´s life. For beef cattle, it is unclear whether changes in metabolic efficiency have been achieved and there are no good early-in-life selection criteria for intake traits suitable for use in selection programs. In practical terms, there is a need for a better understanding of the genetic and phenotypic relationships between feed intake and components of production at different phases in the animal´s productive life.

Both authors are indebted to Meat & Livestock Australia, the Beef CRC and their own organisations for funding.

References

Adam I, Young BA, Nicol AM, Degan AA (1984) Energy cost of eating in cattle given diets of different form. Animal Production 38, 53-56.

Altan O, Oguz I, Akbas Y, Aksit M (2004) Genetic variability of residual feed consumption (RFC) and its relationships with some production traits and fear response in Japanese quail hens (Coturnic coturnix japonica). Archiv fur Geflugelkunde 68, 223-229.

Anon (2005) Handbook of Australian Meat. 7th edn. AUSMEAT, Brisbane.

Archer JA, Barwick SA, Graser, HU (2004) Economic evaluation of beef cattle breeding schemes incorporating performance testing of young bulls for feed intake. Australian Journal of Experimental Agriculture 44, 393-404.

Archer JA, Pitchford WS, Hughes TE, Parnell PF (1998) Genetic and phenotypic relationships between food intake, growth, efficiency and body composition of mice post-weaning and at maturity. Animal Science 67, 171-182.

Archer JA, Reverter A, Herd RM, Johnston DJ, Arthur PF (2002) Genetic variation in feed intake and efficiency of mature beef cows and relationships with post-weaning measurements. Proceedings of the 7th World Congress of Genetics Applied to Livestock Production 10-07.

Archer JA, Richardson EC, Herd RM, Arthur PF (1999) Potential for selection to improve efficiency of feed use in beef cattle: a review. Australian Journal of Agricultural Research 50, 147-61.

Arthur PF, Archer JA, Herd RM, Melville GJ (2001a) Response to selection for net feed intake in beef cattle. Proceedings of the Association for the Advancement of Animal Breeding and Genetics 14, 135-138.

Arthur PF, Archer JA, Johnston DJ, Herd RM, Richardson EC, Parnell PF (2001b) Genetic and phenotypic variance and covariance components for feed intake, feed efficiency, and other postweaning traits in Angus cattle. Journal of Animal Science 79, 2805-2811.

Arthur PF, Herd RM (2008) Residual feed intake in beef cattle. Revista Brasileira de Zootecnia Special Supplement 37, 269-279.

Arthur PF, Herd RM, Basarab JA (2010) The role of cattle genetically efficient in feed utilisation in an Australian carbon trading environment. AFBM Journal 7, 5-13.

Arthur PF, Herd RM, Wilkins JF, Archer JA (2005) Maternal productivity of Angus cows divergently selected for post-weaning residual feed intake. Australian Journal of Experimental Agriculture 45, 985-993.

Arthur PF, Renand G, Krauss D (2001c) Genetic parameters for growth and feed efficiency in weaner versus yearling Charolais bulls. Australian Journal of Agricultural Research 52, 471-476.

Basarab JA, McCartney D, Okine EK, Baron VS (2007) Relationship between progeny residual feed intake and dam productivity traits. Canadian Journal of Animal Science 87, 489-502.

Bordas A, Merat P, Coquerelle G, Noe JP (1995) Influence d´un aliment dilue sur des lignees de poules pondeuses selectionnees sur la consommation alimentaire residuelle. Genetics Selection Evolution 27, 299-304.

Braastad BO, Katle J (1989) Behavioural differences between laying populations selected for high and low efficiency of food utilisation. British Poultry Science 30, 533-544.

BREEDPLAN (2010) BREEDPLAN EBVs "The Traits Explained". ABRI, Armidale.

Bünger L, MacLeod MG, Wallace CA, Hill WG (1998) Direct and correlated effects of selection for food intake corrected for body weight in the adult mouse. Proceedings of the 6th World Congress of Genetics Applied to Livestock Production 26, 97-100.

Byerly TC (1941) Feed and other costs of producing market eggs. Bulletin A1 (Technical). The University of Maryland Agricultural Experiment Station, College Park.

Dantzer R (1981) Physiological aspects of stress in farm animals: Implications for large scale intensive husbandry systems. In: International Symposium of Applied Ethology, Godollo, Hungary. pp. 53-73.

de Haer LCM, Luiting P, Aarts HLM (1993) Relations among individual (residual) feed intake, growthperformance and feed intake pattern of growingpigs in group housing. Livestock Production Science 36, 233-253.

Dobos R, Herd RM (2008) Spectral analysis of feeding patterns of steers divergent in residual feed intake. Australian Journal of Experimental Agriculture 48, 843-846.

Donoghue KA, Arthur PF, Wilkins JF, Herd RM (2011) Onset of puberty and early-life reproduction in Angus females divergently selected for postweaning residual feed intake. Animal Production Science 51, 183-190. 

Egarr AR, Pitchford WS, Bottema MJ, Herd RM, Siddell JP, Thompson JM, Bottema CDK (2009) Fat distribution in Angus steers is related to residual feed intake estimated breeding value. Proceedings of the Association for the Advancement of Animal Breeding and Genetics 18, 536-539.

Falconer DS (1973) Replicated selection for body weight in mice. Genetical Research 22, 291-321.

Foster WH, Kilpatrick DJ, Heaney IH (1983) Genetic variation in the efficiency of energy utilization by the fattening pig. Animal Production 37, 387-393.

Gabarrou JF, Geraert P, Picard, M (1997) Diet-induced thermogenesis in cockerels is modulated by genetic selection for high or low residual feed intake. Journal of Nutrition 127, 2371-2376.

Gilbert H, Bidanel J-P, Gruand J, Caritez J-C, Billon Y, Guillouet P, Lagant H, Noblet J, Sellier P (2007) Genetic parameters for residual feed intake in growing pigs, with emphasis on genetic relationships with carcass and meat quality traits. Journal of Animal Science 85, 3182-3188.

Grobet L, Martin LJR, Poncelet D, Priottin D, Brouwers B, Riquet J, Schoeberlein A, Dunner S, Ménissier F, Massabanda J, Fries R, Hanset R, Georges M (1997) A deletion in the bovine myostatin gene causes the double-muscle phenotype in cattle. Nature Genetics 17, 71-74.

Herd RM (2008) Residual feed intake. In: Resource Allocation Theory Applied to Farm Animal Production, (ed. Rauw WM) pp. 89-109. CAB International, Cambridge.

Herd RM, Archer JA, Arthur PF (2003a) Reducing the cost of beef production through genetic improvement in residual feed intake: Opportunity and challenges to application. Journal of Animal Science 81, E9-E17.

Herd RM, Archer JA, Arthur PF (2003b) Selection for low post-weaning residual feed intake improves feed efficiency of steers in the feedlot. Proceedings of the Association for the Advancement of Animal Breeding and Genetics 15, 310-313.

Herd RM, Arthur PF (2009) Physiological basis for residual feed intake. Journal of Animal Science 87 (E-supplement 14), E64-E71.

Herd RM, Arthur PF, Archer JA (2011) Associations between residual feed intake on ad-libitum, pasture and restricted feeding in Angus cows. Proceedings of the Association for Advancement of Animal Breeding and Genetics. 19, in press.

Herd RM, Arthur PF, Hegarty RS, Archer JA (2002a) Potential to reduce greenhouse gas emissions from beef production by selection for reduced residual feed intake. Proceedings of the 7th World Congress on Genetics Applied to Livestock Production 10-22.

Herd RM, Bishop SC (2000) Genetic variation in residual feed intake and its association with other production traits in British Hereford cattle. Livestock Production Science 63, 111-119. 

Herd RM, Hegarty RS, Dicker RW, Archer JA, Arthur PF (2002b) Selection for residual feed intake improves feed conversion in steers at pasture. Animal Production in Australia 24, 85-88.

Herd RM, Oddy VH, Lee GJ (1993) Effect of divergent selection for weaning weight on live weight and wool growth responses to feed intake in Merino ewes. Australian Journal of Experimental Agriculture 33, 699-705.

Herd RM, Oddy VH, Richardson EC (2004) Biological basis for variation in residual feed intake in beef cattle. 1. Review of potential mechanisms. Australian Journal of Experimental Agriculture 44, 423-430.

Herd RM, Piper S, Thompson JM, Arthur PF, McCorkell B, Dibley KCP (2009) Benefits of genetic superiority in residual feed intake in a large commercial feedlot. Proceedings of the Association for the Advancement of Animal Breeding and Genetics 18, 476-479.

Hoque MA, Kadowaki H, Shibata T, Oikawa T, Suzuki K (2007) Genetic parameters for measures of the efficiency of gain of boars and the genetic relationships with its component traits in Duroc pigs. Journal of Animal Science 85, 1873-1879.

Katle J (1991) Selection for efficiency of food utilisation in laying hens: causal factors for variation in residual food consumption. British Poultry Science 32, 955-969.

Knott SA, Cummins LJ, Dunshea FR, Leury BJ (2008) Rams with poor feed efficiency are highly responsive to an exogenous adrenocorticotropin hormone (ACTH) challenge. Domestic Animal Endocrinology 34, 261-268.

Koch RM, Swiger LA, Chambers D, Gregory KE (1963) Efficiency of feed use in beef cattle. Journal of Animal Science 22, 486-494.

Lines DS, Wolcott ML, Pitchford WS, Bottema CDK, Herd RM, Oddy VH (2009) Some consequences of selection for residual feed intake in beef cattle. Proceedings of the Association for the Advancement of Animal Breeding and Genetics 18, 604-607.

Luiting P (1990) Genetic variation of energy partitioning in laying hens: causes of variation in residual feed consumption. World´s Poultry Science Journal 46,133-152.

Luiting P, Schrama JW, van Der Hel W, Urff EM, van Boekholt PGJJ, van Den Elsen EMW, Verstegen MWA (1991) Metabolic differences between white leghorns selected for high and low residual feed consumption. In: Energy Metabolism of Farm Animals, (eds. Wenk C, Boessinger M.) pp. 384-387. European Association for Animal Production, Kartause.

Luiting P, Urff EM (1991) Residual feed consumption in laying hens. 2. Genetic variation and correlations. Poultry Science 70, 1663-1672.

Luiting P, Urff EM, Verstegen MWA (1994) Betweenanimal variation in biological efficiency as related to residual feed consumption. Netherlands Journal of Agricultural Science 42, 59-67.

McDonagh MD, Herd RM, Richardson EC, Oddy VH, Archer JA, Arthur PF (2001) Meat quality and the calpain system of feedlot steers following a single generation of divergent selection for residual feed intake. Australian Journal of Experimental Agriculture 41, 1013-21.

Montanholi YR, Swanson KC, Palme R, Schenkel FS, McBride BW, Lu D, Miller SP (2010) Assessing feed efficiency in beef steers through feeding behaviour, infrared thermography and glucocorticoids. Animal 4, 692-701.

Mousel MR, Stroup WW, Nielsen MK (2001) Locomotor activity, core body temperature, and circadian rhythms in mice selected for high and low heat loss. Journal of Animal Science 79, 861-868.

Nkrumah JD, Basarab JA, Wang Z, Li C, Price MA, Okine EK, Crews DH, Moore SS (2007) Genetic and phenotypic relationships of feed intake and measures of efficiency with growth and carcass merit of in beef cattle. Journal of Animal Science 85, 2711-2720. Oddy VH (1993) Final Report Project DAN.33 "Increasing the efficiency of muscle growth". Meat. Research Council, Sydney.

Oddy VH (1999) Genetic variation in protein metabolism and implications for variation in efficiency of growth. Recent Advances in Animal Nutrition in Australia 12, 23-29.

Pitchford WS (2004) Genetic improvement of feed efficiency of beef cattle: what lessons can be learnt from other species? Australian Journal of Experimental Agriculture 44, 371-382.

Rauw WM (1998) Undesirable side effects of selection for high production efficiency in farm animals: a review. Livestock Production Science 56, 15-33.

Richardson EC (2003) Biological basis for variation in residual feed intake in beef cattle. PhD Thesis, University of New England, Armidale.

Richardson EC, Herd RM (2004) Biological basis for variation in residual feed intake in beef cattle. 2. Synthesis of results following divergent selection. Australian Journal of Experimental Agriculture 44 431-440.

Richardson EC, Herd RM, Arthur PF, Wright J, Xu G, Dibley K, Oddy VH (1996) Possible physiological indicators for net feed conversion efficiency in beef cattle. Proceedings of the Australian Society of Animal Production 21, 103-106.

Richardson EC, Herd RM, Oddy VH, Thompson JM, Archer JA, Arthur PF (2001) Body composition and implications for heat production of Angus steer progeny of parents selected for and against residual feed intake. Australian Journal of Experimental Agriculture 41, 1065-1072.

Richardson EC, Kilgour RJ, Archer JA, Herd RM (1999) Pedometers measure differences in activity in bulls selected for high or low net feed efficiency. Proceedings of the Australian Society for the Study of Animal Behaviour 26, 16.

Roberts AJ, Paisley SI, Geary TW, Grings EE, Waterman RC, MacNeil MD (2007) Effects of restricted feeding of beef heifers during the post-weaning period on growth, efficiency, and ultrasound carcass characteristics. Journal of Animal Science 85, 2740-2745.

Robinson DL, Oddy VH (2004) Genetic parameters for feed efficiency, fatness, muscle area and feeding behaviour of feedlot finished beef cattle. Livestock Production Science 90, 255-270.

Romans JR, Costello WJ, Carlson CW, Greaser ML, Jones KW (1994). The meat we eat. Interstate Publishers Inc., Danville.

Schenkel FS, Miller SP, Wilton JW (2004) Genetic parameters and breed differences for feed efficiency growth and body composition traits of young beef bulls. Canadian Journal of Animal Science 84, 177-185.

Silverstein JT (2006) Relationships among feed intake, feed efficiency, and growth in juvenile rainbow trout. North American Journal of Aquaculture 68, 168-175.

Tixier-Boichard M, Bordas A, Renand G, Bidanel JP (2002) Residual food consumption as a tool to unravel genetic components of food intake. Proceedings of the 7th World Congress of Genetics Applied to Livestock Production10-06.

Veerkamp RF, Emmans GC, Cromie AR, Simm G (1995) Variance components for residual feed intake in dairy cows. Livestock Production Science 41, 111-120.

Zhuchaev KV, Knyazev S, Hart V, Ephanova N (1996) Associations between stress susceptibility and immune status in pigs. In: Advances in Swine in Biomedical Research Volume 1, (eds. Tumbleson ME, Schook LB) pp. 359-363. Plenum Press, New York.