Measurement of feed intake and feed efficiency in feedlot cattle.

Abstract

Proc. Assoc. Advmt. Anim. Breed Genet. Vol12 MEASUREMENT OF FEED INTAKE AND FEED EFFICIENCY IN FEEDLOT CATTLE. D.L. Robinson,* J.W.Skerrit@ and V.H.0ddy2 *Animal Genetics and Breeding Unit, *CRC for the Cattle and Beef Industry, University of New England, Armidale, NSW 235 1 SUMMARY Measurement of feed intake is repeatable even over short periods, but estimates of weight gain not. Given measurements of intake over a relatively short period (7-9 weeks including time adjust to the automatic feeding system), a more reliable assessment of efficiency was achieved relating feed intake to gain modelled using a growth curve based on all weights recorded in feedlot, instead of simply using weights for the period when intake was recorded. Keywords: Beef cattle, feed intake, weight gain, feed efficiency are to by the INTRODUCTION Feed is a major cost in the production of grain-fed beef. Ideal feedlot stock should utilise feed efftciently and economically, ie grow well with moderate feed consumption. If efficiency can be improved, benefits may extend beyond the feedlot. It is thought that animals which grow efficiently on grain could also grow efficiently on grass and perhaps have lower feed requirements when mature. If so, identification of efficient feedlot stock might reduce production costs not just for the feedlot, but also the breeding herd, estimated to account for 5289% of metabolizable energy needed for beef production (Thompson and Barlow 1986). Accurate measurement of feed efficiency, however, requires a testing period of 10 or more weeks (Archer 1996). The main expenses relate to measurement of intake, which may be repeatably measured over shorter periods, but at least 10 weeks are required for accurate measurement of weight gain. Here we consider consequences on accuracy of selection and effects for other traits of relating a more limited set of intake data to growth modelled from all weights recorded in the feedlot. MATERIALS AND METHODS The design of the Cooperative Research Centre for the Cattle and Beef Industry (CRC) research programme was described by Robinson (1995). Measurements analysed here were of 308 steers from two groups of Bos tuurus steers purchased at weaning in 1995 and 1996 from 9 Angus, 4 Shorthorn, 3 Hereford and 2 Murray Grey herds and a group of Bos indicus steers purchased in July 1994 from a Santa Gertrudis and a Brahman herd. Taurus steers were grown out at the Agricultural Research Station, Glen Irmes on pasture, or pasture supplemented with concentrates or forage in 1995 and in 1996 on the same pastures, unsupplemented, but subject to differences in pasture fertility and possible carry-over effects from 1995. Zndicus steers were grown out in central Queensland until December 1994, then transferred south for a further three month's grow-out on pasture. Feedlot finishing commenced once each group averaged 400 kg. Groups were further subdivided for finishing to Korean (target 520 kg) and Japanese (640 kg) markets, making a total of six management groups or cohorts. A specially written suite of computer programs was used to find the most efficient experimental design, given the number of breeds, birth herds, numbers of offspring of each sire, desired treatment combinations and genetic links with other 287 Proc. Assoc. Advmt. Anim. Breed Genet. Vol12 groups/years/treatment combinations. Table 1 shows numbers of animals, breeds and birth herds for the six cohorts, plus mean feed intakes, weights and weight gains. Table 1. Numbers of steers, measurements, and means by group (I=indicus, T=tuurus) Numbers of Group & Market 194K 1945 T94K T94J T95K T95J weeks in AF pens 8 12 7 9 8 9 Weight records In AF Topens ta1 4 9 7 15 4 9 5 13 5 8 5 11 Means while in AF (automatic feeder) pens Gain (kg/ day) 1.21 0.75 1.26 1.12 1.90 1.51 Intake (kg/ day) 12.10 10.29 11.73 11.36 14.56 14.29 eating sessions (no./day) 12.4 14.2 4.7 5.5 6.0 7.1 Time feeding (min/day) 89.1 81.5 103.0 90.6 122.0 107.4 Steers 16 17 81 82 55 57 Birth herds 2 2 11 11 7 7 Weight (kg) 494 552 499 575 499 525 After introductory and intermediate diets in weeks 1 and 2, steers were fed standard finisher rations. Ad libitum intake was measured by automated feeders (AF), developed by the CRC. Each group pen has a feeding space which only one animal may access at a time. An electronic identification tag is read every time an animal enters the feeder and weight of food plus time spent eating recorded. Due to limited capacity of AF pens, intake was measured first on Korean steers. Japanese steers were measured at a later part of their growth curve, with heavier average weights and lower weight gains (Figure 1). Weights were recorded approximately fortnightly throughout time in the feedlot, resulting in 8-9 measurements for Korean steers and 11-15 for Japanese. Table 2 shows pooled correlations of gains from the first 8 weighings (14 weeks) in the feedlot. Weight gains were not repeatable, even over longer periods. Correlations between gains in weeks 1-8 and weeks 9-16 were 0.71, 0.43, -0.09, .02, -0.07, -0.16 for cohorts 194K, 194J, T94K, T94J, T95K, T95J respectively. Table 2. Correlations (pooled over cohorts) of successive weight gains (from approximately fortnightly measurements, below diagonal) and weekly intake measurements (above diagonal) Intk I 0.561 Gain Gain Gain Gain Gain Gain 1 2 3 4 5 6 -0.388 0.008 0.130 0.036 -0.009 -0.053 Gain 2 -0.274 0.000 0.088 0.000 -0.002 Gain 3 Intk 2 0.434 0.704 -0.334 0.065 -0.034 0.106 Gain 4 Intk 3 0.318 0.511 0.600 -0.3 10 0.085 -0.130 Gain 5 Intk 4 0.323 0.500 0.627 0.509 -0.344 0.128 Gain 6 Intk 5 0.375 0.571 0.650 0.586 0.710 -0:374 Gain 7 Intk 6 0.317 0.584 0.546 0.481 0.502 0.671 Intk Intk Intk Intk Intk Intk 2 3 4 5 6 7 288 Proc. Assoc. Advmt. Anim. Breed Genet. Vol1.2 Figure 1. Fitted growth curves by nutrition for 4 cohorts of taurus steers (lines), together with raw means (points) and entry/exit to automatic feeder (AF) pens (arrows). (R= pasture (P), A= P + concentrates, 0 - P + forage; supplements 1995; pasture fertilitykarry-over ef'fM.s 1996). Cohort TMK C#wt 530 . . T95K Statistical modelling of weight &cords. FOr 6ach cohort, th' wei,+ht,w+ of tier i at time j, mod&e@ using.prOgmm ASIUML (Gilmour 199QJby .the equatb WV = p + iclj +\qdj' + abi + hi + Pi + t+ + II&j * Rqidj' 8i + &dj + @j2 + % `4 + Wbak?d- repW8entsiday aumber in the f&240$, C&d so day 1 Wa day 0-f entrjr *Ihe Was (1) AF ,peAs. equation, p f Mb + qd.2, thm&m I&O&& the ovwaI4 gmwth ; aa, rllQd& I$ + 3 + W# `Z4WdWI &lBg@Liil the bi?&`kfdi&lld &y 02 weighing33*, 3. growth paBe& due 0 bacitgrounding tmtritkm; ai c pidJ + ~4 de&ions in the gmw&~patkrn for animal i. All terms except the first four ware assumedPto be r&ndom efWts; so the magnitude of the fitted values reflected the variation of each term. The two error terms, eij and Ed, represent the two kinds of error found in sequences of measurem$nts such, as ~@&tsT T&, @st @.a r&Oqt ermr, urb~@tte+I with errors at Other *es, due to v++l$iiy in the Hieight-dfi fbe a$&l oyer the course Of a day, ,ZWV,~II,,~eight. q;l~~n~ts apd.qtha +oourcesOf yariati?? svh ,w,naeUrati,sns fkm thy *at6 @-pial .feeding pattern. The secoad represents longer,*pm .depw@pzstic@ a typic4 grq&h pat#err~:~h as illness or cha&es ia eat&g, heha&kdrhkh *Gt`mOri t&$Sne weight rec;ord- The latter error term was model&54by a.&@ Order autoregressive *prok2zss [C!hat@ld 1975). *:`hi 8ld.p The qkratic 289 Proc. Assoc. Advmt. Anim. Breed. Genet. Voll2 Cohort 194K 194J T94K T94J T95K Correlations with feed intake Metabolic wt 0.76 0.84 0.51 0.60 0.81 GNm (model) 0.95 0.94 0.59 0.82 0.76 GNa (actual) 0.82 0.80 0.30 0.69 0.57 RFIm (model) 0.3 1 0.35 0.75 0.52 0.43 RFIa (actual) 0.47 0.49 0.81 0.56 0.46 Correlations with GNm (modelledfrom whole time in feedlor) RFIm (model 0.00 0.00 0.00 0.00 0.00 RFIa (actual) 0.23 0.29 0.29 0.26 0.21 Residual variation in intake (kgz) after fitting metabolic weight GNm (model) 0.48 0.51 1.03 0.75 0.57 GNa (actual) 1.10 1.02 1.19 0.85 0.67 Partial regression coefficients for GNm imodel) 6.42 5.95 3.97 6.33 3.31 GNa (actual) 2.92 2.23 1.43 2.91 1.97 T95J 0.68 0.72 0.58 0.48 0.53 0.00 0.24 plus 0.79 0.97 4.43 2.36 Table 3. Correlations of mean feed intake and metabolic weight with modelled and actual weight gain. RF1 estimated from above, by cohort (I=indicus, T=taurus) RESULTS AND DISCUSSION Growth curves. Overall, the model represented the data well. Figure 1 shows fitted curves by nutrition for taurus cohorts together with observed means and times of entry/exit to AF pens. Growth of steers up to Korean market weights was approximately linear with a small amount of curvature towards Japanese market weights. Some compensatory gain was evident in T95J. On any given day, means for each nutrition group show a similar effect, either all being above, below,or on the fitted curves. Entry/exit to AF pens had, however, no obvious effect on growth, suggesting current management practice of allowing up to 11 steers per AF pen was not a limiting factor. All cohorts show relatively large variation between animals and in three cohorts there was also substantial variation in weight due to breed/birth herd. The residual variation was of similar magnitude, 75-98 kg2 in all cohorts. The autoregressive error term was highly significant in all cases, being . l l -.45 the size of the residual variation, eij. Relationship with feed intake. Residual feed intake was calculated as the residuals of the regression of average daily feed intake on weigh$.73 and average daily gain, either actual (GNa, giving residual intake RFIa), or modelled (GNm, RFIm). Modelled gain included all terms in equation 1 except dates of weighing, pj and the error terms eij and Ei'.Table 3 shows the residual variation was smaller in every cohort when GNm was used instead o! GNa, indicating substantially more of the variation in feed intake could be explained by GNm. For indicus cattle the correlations between GNm, and intake were 0.95 and 0.94 for Korean and Japanese markets implying that gain was very strongly related to intake. Steers choosing to eat more, gained weight faster in the feedlot. The 290 Proc. Assoc. Advmt. Anim. Breed Genet. Vol12 indicus cattle tested here also had different eating patterns, with more but shorter feeding sessions per day (Table 1). For this group of cattle, the data indicate it would be difftcult to reduce voluntary intake of food without also reducing weight gain. For tuurus cattle, correlations between intake and GNm were lower, allowing some scope for reducing intake without lowering gain. In general, RFIa was more closely related to intake than RFIm. Furthermore, though RFIa was not correlated with gain in the AF feeders, it was correlated with RFIm. This is because RFIa is based on an inaccurate measure of gain, so adjustment for gain is incomplete, as indicated by the relatively low partial regression coefficient for GNa (Table 3). The incomplete adjustment means that animals with lowest RFIa will tend to have lower weight gains and care will be required to prevent selection for RFIa reducing growth. Accurate assessment of efficiency is not limited by measurement of food intake, but by measurement of weight gain. Modelled gain, eg GNm, should increase efficiency of selection relative to actual gain. REFERENCES Archer, J. ( 1996) PhD Thesis, University of Adelaide. Chatfield, C. (1975) 'The analysis of time series: theory and practice'. Chapman & Hall, London. Gihuour, A. (1996) ASRBML Manual. NSW Agriculture, Orange, NSW 2800, Australia. Robinson, D.L. (1995) Proc. Aust. Assoc. Anim. Breed Genet. 11541. Thompson, J.M. and Barlow, R. (1986) Proc 3rd World Cong. Genet. Appl. Live& Prod. X1:271. 291