J. Anim Sci. 2006. 84:1778-1789. doi:10.2527/jas.2005-514
© 2006 American Society of Animal Science
Body fatness affects feed intake of sheep at a given body weight1,2
B. J. Tolkamp*,3,
G. C. Emmans* and
I. Kyriazakis*,
* Animal Nutrition and Health Department, SAC, West Mains Road, Edinburgh EH9 3JG, UK;
and
Veterinary Faculty, University of Thessaly, PO Box 199, 43100 Karditsa, Greece
 |
Abstract
|
|---|
In a 1-yr experiment, nutritional treatments were used to produce different combinations of BW and BCS in lambs. The experiment served to quantify the effects of BW and BCS on ADFI by sheep. Ewe lambs (n = 78) were assigned to treatment groups that had ad libitum access to one feed at a time. Three feeds were used: a medium-quality chopped hay (L), a pelleted feed based on oat feed (M), and a pelleted feed based on barley (H). Three groups received only one of these feeds throughout. Two groups first received H and then were switched to M when they reached a BW of 45 or 65 kg. Two groups first received L and then were switched to M or H after reaching a BW of 45 kg. Three groups first received H or M but were switched to L after reaching a BW of 45, 65, or 95 kg. Daily feed intake, BW, and BCS were recorded, and ME content of the feeds was estimated in a separate digestibility experiment. The lambs consuming M ate more (P < 0.001) feed than lambs consuming H, but this had no significant effects on ME intake or gain in BW or BCS. Animals that had had access to L were lean for their BW when switched to H or M and showed compensatory intake and gain. Animals switched from M or H to L all lost BCS; BW change depended on the BW at the switch. The treatments produced different combinations of BW and BCS for animals with access to the same feed. The ADFI of a given feed varied systematically with BCS for animals of a given BW. The model ADFI = a x BW x [1 – (b x BCS)] gave a reasonable description of the data in all treatments. A model using BW, BCS, and their interaction gave a slightly better fit but explained little more of the variation in ADFI than the simpler model. The implications of the collected data are that BW alone is an insufficient descriptor of the animal to correctly predict feed intake and that intake predictions can be improved by taking BCS into account.
Key Words: body condition score • body weight • feed intake • modeling • sheep
|
INTRODUCTION
|
|---|
In principle, sufficient descriptions of the animal, the feed, and the environment allow ADFI to be predicted (Emmans and Kyriazakis, 2001
). For ruminants it is generally assumed that only animal and feed factors have to be included in predictive models, as long as extremes in the physical environment are avoided and animals are healthy, housed individually, and under fixed day length (ARC 1980; NRC 1987 NRC 2000). There is extensive literature on the effects of feed quality on ADFI by sheep of a given BW, but much less work has been done on the effects on ADFI of variation in the state, especially the fatness, of a nonreproducing sheep of a given BW.
The regression equations of ARC (1980) predict ADFI from BW and feed quality, and these are incorporated in the NRC (1987) ADFI predictions for sheep. In these equations, BW is the only animal variable that affects ADFI. This is also generally the case in other models that predict ADFI by sheep (Pittroff and Kothmann, 2001). However, the fatness of a given genotype can affect ADFI (Foot, 1972; Sibbald and Rhind, 1997). Despite this, no models are available that can take account of a sheep’s actual fatness on intake.
There is a need to develop predictive models that take account of the effects of animal fatness on ADFI. The most widely available estimator of fatness in live sheep is at present BCS (Russel et al., 1969). The experiment described here was intended to quantify the effects of variation in BW and BCS on ADFI by sheep consuming feeds of different quality. The experimental data are presented, and the suitability of some simple models to describe these results was tested.
|
MATERIALS AND METHODS
|
|---|
The Animal Experiments Committee of the Scottish Agricultural College approved the protocol that was used in the experiment described.
Animals, Feeds, and Housing
Seventy-eight female Greyface x Texel crossbred lambs, born in April and May 2000, were weaned at 12 wk of age and received access to hay and concentrates until they were brought into the experimental facilities in the first week of August. Pens were constructed with mesh bottoms, and the lambs were housed individually, separated by metal tubes. This pen construction allowed the lambs to see, hear, and smell one another, whereas ADFI could be measured individually. To avoid effects of day length on intake and performance, high intensity artificial light was provided for 18 h daily throughout the experiment.
After an adaptation period to the facilities of 1 wk, lambs were assigned to treatment groups. The experiment lasted 50 wk, during which lambs had ad libitum access to 1 of the 3 feeds used at any one time. The lowest quality feed was chopped grass hay (L, predominantly Lolium perenne), and the medium (M) and high (H) quality feeds were pelleted. The ingredient compositions of M and H and the chemical compositions of all feeds are given in Table 1. It was intended that growth rate would decrease in the order H > M > L.
Individual BW, to the nearest 0.1 kg, and amounts of feed offered and refused, to the nearest gram, were recorded weekly. Body weight and feed refusals were not recorded at the end of wk 1 (adaptation week) or at the end of wk 29 (as a result of a misunderstanding). Wool was clipped in wk 39 or 40 and weighed. Body weight values recorded after clipping were increased by the measured weight of the removed wool, which averaged 5.9% of BW before clipping. To determine the DM content, daily samples of feed offered were pooled in weekly batches to be dried at 95°C for 23 h. These weekly samples were later pooled for chemical analyses. At approximately 2-wk intervals throughout the experiment, the same person recorded BCS to the nearest quarter unit on a scale from 0 (emaciated) to 5 (extremely fat), as described by Russel et al. (1969).
Experimental Design and Data Analyses
Lambs were assigned to 9 treatment groups. Three groups (treatments L, M, and H) each consisted of 10 lambs and were fed the single feeds described in Table 1
throughout. The other 6 treatment groups consisted of 8 lambs each. Two treatments (H45M and H65M) received H initially and were then switched to M after reaching 45 or 65 kg of BW, respectively. These treatments were chosen to obtain data on intake of M and performance of lambs with high BCS for BW. Two further treatments (H45L and H65L) initially received H but were switched to hay after reaching 45 and 65 kg of BW, respectively. These treatments were designed to measure hay intake by lambs with greater BCS values for BW than those on L continuously. Two further treatments (L45M and L45H) initially received L but were switched to M and H, respectively, after reaching 45 kg of BW. These treatments were chosen to measure the intake of the greater-quality feeds by lambs with low BCS values for BW compared with those fed M or H continuously.
In wk 27, one animal from treatment H45M died, and the postmortem suggested that the cause was Cu poisoning. Subsequently, blood from all lambs was sampled, and the plasma was analyzed for Cu and aspartate aminotransferase. Several individuals fed M or H (but none on L) showed high levels of Cu (>18 µmol/L) or aspartate aminotransferase (>400 IU/L) or both. Consequently, all sheep on M or H were treated with molybdate (Humphries et al., 1988). The mineral composition of M and H was changed from wk 30 of the experiment onward by adding 2.5 kg of Copper Check Premix (Frank Wright, Ltd., Derbyshire, UK) per 997.5 kg of pellets; this added Fe (as ferrous sulfate), Mo (as sodium molybdate), S, and Ca. Nevertheless, 2 more lambs with signs of Cu poisoning had to be removed between wk 27 and 30 of the experiment. The data for 5 additional lambs that completed the experiment were removed from the data set before analysis because ADFI and BW gain showed a clear decrease during the period that Cu poisoning was occurring. Decreases in intake and growth are generally observed only after Cu levels in the body become toxic and before lambs die (Lippi et al., 2003), and we are confident that intake was not affected by Cu toxicity in the data that were analyzed. Apart from the loss of data as a result of Cu poisoning, only 1 ewe (from treatment L) was removed (with veterinary advice) for general weakness, and her data are not included in any analysis. The treatments and final number of lambs used in the analyses are given in Table 2.
View this table:
[in this window]
[in a new window]
|
Table 2. Overview of treatments and number of lambs that were originally assigned to treatments, that completed the experiment, and that were included in the analyses
|
|
The experiment departed from that intended in 2 ways. The first departure was that by experimental wk 40 many of the lambs receiving the H or M feeds had reached a BW of more than 90 kg and a BCS of 5. Some individuals began to show the first signs of leg problems that were attributed to the high BW. Lambs in treatments with a mean BW of more than 90 kg were therefore removed from their treatments for health and welfare reasons beginning in wk 41. A number of lambs with no obvious leg problems in group M were switched to L for the last 10 wk of the experiment to obtain data on the hay intake of very fat lambs; the others were removed from the experiment altogether. The second departure arose from the intention that feeds M and H would result in different rates of gain in BW and BCS. During the experiment, it became clear that lambs receiving M and H showed a similar performance. It was, therefore, necessary to test first whether the feeding of M and H led to lambs that differed in performance before further analyses could be carried out.
The first analysis used the data of treatments H, H45M, H65M, and M up to wk 40 only (i.e., before any lamb was removed because of fatness). We used these data to establish whether feeding of M led to differences in either BW or BCS compared with feeding H. The same subset of data was used to analyze the effects of feed quality on ADFI and intake of ME.
The second analysis used the data of treatments H, M, L45H, and L45M only to test the hypothesis that BCS would affect ADFI by lambs fed M or H. For part of this analysis, experimental weeks were redefined in terms of synchronized weeks; synchronized wk 1 was defined as the first week after an individual reached 45 kg of BW. For lambs in treatments L45M and L45H, synchronized wk 1 coincided with the first week in which an individual began receiving M or H instead of hay.
The third analysis used the data of treatments L, H45L, H65L, and of lambs fed M up to wk 40 but hay thereafter (M95L) only to test for the effect of BW and BCS on hay intake and performance. Finally, several models describing ADFI of the 3 feeds on the basis of BW and BCS were parameterized using data of all treatment groups.
Measurement of Digestibility
Alongside the main experiment, 12 additional Greyface x Texel cross ewe lambs born in the same period in the same flock were used to measure the digestibility of the 3 experimental feeds. The BW of these lambs during the digestibility experiment was 49.0 (SD = 6.1) kg. No BCS of these lambs was taken. Four lambs each received one of the experimental feeds ad libitum for 3 wk while they were housed in metabolism cages. Feed offered, feed refused, and feces produced were measured to the nearest gram during the last week of this period. Samples of offered feeds, feed refusals, and feces were analyzed for DM (drying at 95°C for 23 h), ash (550°C for 24 h), and GE (by bomb calorimetry). From these data, the ME yields of the experimental feeds were estimated as 0.82 x DE contents (NRC, 1985).
Statistical Analyses
Data were screened for errors before any analysis took place. The raw data were used for some of the analyses. Where the data of individuals were synchronized as explained earlier, missing values were estimated. Missing BW in wk 29 were interpolated linearly from BW in wk 28 and 30. The BW of one ewe was missing for 1 wk, and this value was similarly interpolated. From the amounts of feed offered in wk 29 and 30 and feed refusals recorded at the end of wk 30, the intake during wk 29 and 30 was calculated. Subsequently, it was assumed that lambs had consumed half of this amount in each of wk 29 and 30. Body condition score was smoothed using the 4253H twice algorithm (Velleman, 1980) using Minitab Release 12.1 (Minitab Inc., State College, PA).
When lambs were switched during the experiment from hay to pellets or vice versa, smoothing was carried out separately for the 2 periods. For the first period, all observations obtained in that period were used in the smoothing process. For the second period, all observations in that period plus the last observation from the first period were used in the smoothing process. Subsequently, missing values were interpolated linearly from the smoothed data sets. The data sets of BCS thus obtained were used in all analyses, where data obtained with individuals were synchronized, and for the modeling part of the study. The first 2 weekly observations of lambs receiving a novel feed were considered part of the adaptation period and these were therefore excluded from all statistical analyses.
Analysis of variance was used to test differences between treatment groups. The nonlinear model (see discussion for a justification):
 | [1] |
was fitted using the Gauss-Newton optimization procedure of GenStat (Lawes Agricultural Trust, 2004). Feed intake was expressed as kilograms per day. First, parameters ai and bi were estimated per feed (with i = L, M, or H). Subsequently, the model: ADFI = ai x BW x [1 – (b x BCS)] was fitted across feeds, where parameter ai was estimated per feed as before, but parameter b was estimated pooled across feeds. An F-test was used to analyze the effect of estimating parameter b within vs. across feeds. The term relative feed intake (RFI, g·kg–1·d–1) was used whenever feed intake was expressed relative to BW.
Linear regression was used to test whether variation in ADFI for the 3 feeds separately could be described significantly better by other combinations of different expressions of BW and BCS. The full model was:
 | [2] |
In succession, we first dropped (b6 x BW2 x BCS2 + b7 x BW x BCS2 + b8 x BW2 x BCS), to give the model:
 | [2a] |
and then dropped (b4 x BW2+ b5 x BCS2), to give the model:
 | [2b] |
Finally, a model was fitted without an intercept:
 | [2c] |
|
RESULTS
|
|---|
Digestibility Experiment
Mean intakes of fresh feed during the digestibility experiment were 32 (SD = 0.8), 64 (SD = 2.9) and 46 (SD = 6.2) g/kg of BW for feeds L, M, and H, respectively. The digestibility of the hay estimated with 1 of the 4 lambs was anomalous, and the estimated ME yield of the fresh hay (7.9, SD = 0.45, MJ/kg) was therefore based on the observations of 3 lambs only. The ME yields of H and M pellets were estimated at 10.2 (SD = 0.66) and 8.5 (SD = 0.18) MJ/kg.
Intake and Performance of Lambs Fed High or Medium Pelleted Feeds
Figures 1a and 1b show the group mean BW and BCS of the 4 treatments receiving pelleted feeds up to wk 40 of the experiment. Analysis of variance showed that total gains in BW and BCS up to wk 40 were not different among these treatments (P = 0.78 and P = 0.99, respectively) but that ADFI differed among dietary treatment (P < 0.001; Table 3; Figure 1c). Table 3 also shows that the total ME intake and the ME intake per kilogram of BW gain were not significantly affected by treatment (P = 0.21 and P = 0.81, respectively). The ADFI of M by lambs in treatment H45M during wk 15 to 40 and by lambs of treatment H65M during wk 28 to 40 did not differ (P = 0.48 and 0.15, respectively) from the ADFI of M by lambs in treatment M during the same periods. These analyses show, therefore, that treatments H and M did not result in differences in performance (gain in BW or BCS) and that the ADFI of M was unaffected by whether H or M was consumed previously. In the following analyses of the effects of animal state on consumption of hay, lambs previously receiving M or H are therefore treated as lambs with the same nutritional history.
View this table:
[in this window]
[in a new window]
|
Table 3. Intake and performance of lambs receiving feed barley-based pellets (H) or oat-feed-based pellets (M) throughout, or H up to 45 or 65 kg and M afterward (H45M and H65M, respectively)
|
|
Effects of Animal Fatness on Performance and the Intake of Pelleted Feeds
Ewes in treatments L45H and L45M received hay until 45 kg of BW and H or M afterwards. Figure 2 shows the mean BW, BCS and ADFI from 5 wk before until 25 wk after the feed switch in treatments L45M (top graphs) and L45H (bottom graphs). Data obtained in treatments M and H were similarly synchronized with respect to the week in which lambs were observed to have passed 45 kg of BW, even though no change in diet occurred in these treatments. For comparison, the means of treatments M and H after synchronization are shown also in Figure 2. Lambs fed hay gained less BW and BCS in the weeks preceding the feed switch than lambs that received M or H at similar BW (P < 0.05). Lambs gained BW and BCS faster in the weeks after the switch than lambs with similar BW but greater BCS that had received M or H from the beginning of the experiment (Figure 2). This was related to a greater (P < 0.05) intake of H and M by lambs in treatments L45H and L45M than by lambs in treatments H and M, respectively, for most observations between wk 4 and 15 after the switch (Figures 2c and 2f). Between wk 4 and 15 after the switch, the mean ADFI by lambs that received hay previously was greater (P < 0.001) by 22 ± 1.5% compared with lambs that had received M or H from the beginning of the experiment. The proportional increases were 23 ± 2.1% and 21 ± 2.3% for H and M, respectively, which did not differ (P = 0.6) between feeds. None of the intakes differed between treatments H and L45H or between treatments M and L45M between 15 and 25 wk after the switch.
View larger version (23K):
[in this window]
[in a new window]
|
Figure 2. Group mean weekly BW, smoothed BCS, and ADFI against week from reaching 45 kg of BW for treatments M and L45M (top panels) and treatments H and L45H (bottom panels). Open symbols represent the treatments M and H in which lambs received M or H, respectively, throughout the period shown. Closed symbols represent treatments L45M and L45H that received L until wk 0 and M or H, respectively, in wk 1 to 24. L = chopped hay; M = oat-feed-based pellets; and H = barley-based pellets. The residual SD (RSD) is also given.
|
|
Effects of Animal State on Performance and the Intake of Hay
Figure 3 shows mean BW and BCS of lambs that received hay in the latter part of the experiment (individuals from treatments L, H45L, H65L, and M95L). Lambs consuming hay throughout (treatment L) gained BW but lost some BCS during the experiment. Lambs receiving hay after consuming pellets up to 45 kg of BW (treatment H45L) gained BW slower and lost BCS slightly faster than lambs in treatment L. Lambs that switched from pellets to hay at 65 kg (group H65L) almost ceased to gain BW and lost condition rapidly. Lambs that switched from pellets to hay at around 95 kg lost BW and BCS rapidly during the last 10 wk of the experiment.
Figure 3c shows the mean hay intake of treatment groups L, H45L, H65L, and M95L. During the first part of the experiment, lambs from treatments L45H and L45M consumed similar amounts of hay as lambs in treatment L, and their data are not reproduced in Figure 3 to avoid clutter. Groups L and H45L consumed similar amounts of hay from experimental wk 10 even though lambs in group H45L had greater mean BW than lambs in group L. After switching to hay, lambs in treatment H65L initially consumed less than lambs from groups L and H45L, but daily intakes were more similar in these 3 groups toward the end of the experiment. Hay intake by lambs in treatment M95L was less than hay intake by the other 3 groups during the last 10 wk of the experiment. When hay intake was expressed relative to BW (Figure 3d), mean intake decreased in the order L, H45L, H65L, and M95L. Within each treatment, hay intake tended to increase with experimental week (Figure 3d).
Using Models to Describe the Effects of BW and BCS on Intake of Hay and Pellets
Model 1, ADFI = a x BW x (1 – [b x BCS]), where a and b represent regression coefficients, was first fitted to data obtained from all lambs in all weeks. Estimating separate a values for each feed resulted in a large decrease in residual SD (RSD) from 543 to 293 g/d. The estimates (± SE) for a were 39.3 (± 0.3), 86.6 (± 0.9), and 72.1 (± 0.72) for feeds L, M, and H, respectively, and for b pooled across feeds, 0.140 (± 0.0008). The a and b-values were also estimated separately for each feed. An F-test on the basis of residual sums of squares obtained with these 2 models showed that the variation in intake was described significantly better when b-values were estimated per feed. However, this addition decreased the RSD trivially from 293 to 290 g/d and increased the percentage of the variance accounted for only from 77.4 to 77.9%. The resulting estimates for a and b are in Table 4. Figure 4 shows RFI plotted against BCS with the fit of the model for each feed.
View this table:
[in this window]
[in a new window]
|
Table 4. The residual SD (RSD) associated with all models fitted per feed (L, M, and H) and the estimated regression coefficients ± SE for Models 1 and 2c1
|
|
View larger version (8K):
[in this window]
[in a new window]
|
Figure 4. Group-mean daily feed intakes measured weekly relative to BW (RFI) in relation to group mean BCS for treatments consuming L (a), M (b), and H (c). L = chopped hay; M = oat-feed-based pellets; and H = barley-based pellets. The solid lines represent the regression lines derived with model (1); i.e., for L: RFI = 42.2 x (1 – [0.155 x BCS]), for M: RFI = 81.4 x (1 – [0.134 x BCS]), and for H: RFI = 74.3 x (1 – [0.143 x BCS]).
|
|
The various forms of model 2 were fitted to the feeds separately. Model 2c was no worse than any of the more complex models. For all 3 feeds, the coefficients for BW, BCS, and BW x BCS were significant (P < 0.05), and Table 4 shows the RSD and estimated parameters.
|
DISCUSSION
|
|---|
Accurate prediction of feed intake by animals requires a sufficient description of the animal and the feed. There are numerous efforts at predicting the intake of ruminants in relation to feed quality (e.g., ARC, 1980; NRC, 1987; Tolkamp and Ketelaars, 1992), although the most appropriate framework for such ruminant intake models is still contentious (Emmans and Kyriazakis, 1995; Forbes, 1995; Ketelaars and Tolkamp, 1996; NRC, 2001; Kyriazakis, 2003). For a description of the animal, however, the majority of existing models rely on BW only. There are numerous examples that use BW, or metabolic size (BW0.75 or BW0.73), or metabolic size squared, or BW in an exponential equation, or some combination of these expressions (ARC 1980; NRC 1987; CSIRO, 1990; Pittroff and Kothmann, 2001). From the review by Pittroff and Kothmann (2001), it seems that for nonlactating sheep of a given genotype, BW is the only animal characteristic included in 7 out of the 11 models that were analyzed. Pittroff and Kothmann (2001) classified 4 models as including some estimate of body condition. It seems, however, that also in these models the only actual information about the animal that is used is BW. In one model, intake is assumed to decline linearly with BW as the animal matures, whereas the 3 other models combine actual BW with some estimate of normal weight for age or with (potential) mature BW. The use of mature size alone or a description of the "normal weight for age" (for instance, in the form of a Gompertz equation as used by Vera et al., 1977) is potentially very useful, especially if BW is the only information that is available about actual animal state (Emmans, 1997; Lewis et al., 2002). We are, however, not aware of any models that attempt to describe intake on the basis of the 2 main characteristics of growing sheep of a given genotype that may actually be available on-farm: BW and BCS. There are a few reports in the literature of short-term experiments that show a negative effect of animal fatness as measured by BCS on feed intake relative to BW or metabolic size of mature sheep (Foot, 1972; Sibbald and Rhind, 1997). The data set presented here shows clearly that BW alone (in whichever expression used) is insufficient to account for the variation in intake of a given feed. This is most clearly demonstrated for feeds M and H in Figure 2 and for L in Figure 3. These figures show that variation in animal fatness is associated with variation in intake at a given BW in lambs. However, before discussing the attempts to describe the observed effects of sheep fatness on feed intake, we will discuss the experimental results.
Effects of the Quality of Pelleted Feed on Intake and Performance
In the present experiment 3 feeds were used: hay (L), medium quality pellets (M), and high quality pellets (H). The expectation was that the BW gain of animals would increase in the order L, M, and H. Similarly, the expectation was that at a given BW, the BCS would increase in that same order. However, Figure 1 and Table 3 show that performance of the lambs fed H and M did not differ, a problem that has plagued other researchers in the past (Blaxter et al., 1982). This was caused by a more than 20% greater ADFI by lambs consuming M compared with H at similar BW and BCS. This means that the nutritional treatments were essentially limited to a contrast between 2 feed types: hay and the pelleted feeds used.
Some animals reached a BCS of 5 some weeks before experimental wk 40. It seems likely that in these last few weeks animals would have become slightly fatter still (but the scorer could not detect this). Because the increase is likely to have been small and the same happened in all treatments, it was ignored in the analysis. Although condition scoring is a good method to estimate animal fatness under most conditions, the method has its limitations for extremely fat animals.
Effects of Previous Feed on the Intake of Hay
Animals receiving hay from the beginning of the experiment or from 45 kg gained less BW and BCS than animals receiving pellets, as intended. Loss of BCS occurred in all other treatments after animals began to receive hay (Figure 3b). The loss in BCS of animals in treatment L was very small. This was evidently different for animals in treatment H65L. Animals in this group lost BCS rapidly after being switched to hay, even though their BW changed very little. Possibly, some increase in gut fill, associated with the gradual increase in hay intake, may have contributed to BW maintenance, despite lipid losses in this treatment group. In addition, sheep may have gained some lean during this period of lipid loss, but the data we collected do not allow us to distinguish between these possibilities. From the rapid loss in BCS of animals in treatment M95L it is evident that most of the BW loss in this group after wk 40 must have been due to the loss of lipid.
Treatments L, H45L, H65L, and M95L allowed us to measure hay intake by animals with a BW that varied from 35 to more than 90 kg and a BCS that varied between 2 and 5. Mean RFI from hay decreased in the same order that mean BCS increased. In addition, the within-group hay intake increased faster in time when the loss of BCS was faster. This suggests an important role for body fatness in hay intake regulation. Furthermore, Figure 3 suggested a possible relationship between ADFI and BW. Relative intakes of 25 g of hay/kg of BW can be observed in group L at the beginning of the experiment, around wk 30 to 35 in group H45L, and toward the end of the experiment in group H65L (Figure 3d). Figure 3a shows that the mean BW of these groups during those periods is very variable (from around 35 to around 67 kg). However, Figure 3b shows that these groups all have a BCS around 2.5 during the periods that they have RFI of around 25 g of hay/kg of BW. That suggested that hay intake, at least over this BW range, is proportional to BW when animals have the same BCS. This observation led to model (1) to describe the effects of BW and BCS on feed intake.
Effects of Previous Feed on the Intake of Pellets
To compare the intake of group L45M with group M, and of group L45H with group H, the data were synchronized relative to the week animals reached a BW of 45 kg. The reason was that individual animals were switched to the other feed in the week after reaching this target BW. Individuals in treatments that were switched from M or H to hay reached the target BW within weeks of one another because rates of growth were high in animals receiving these feeds. However, growth rates were much lower in lambs receiving L. There was, therefore, a large difference between individuals in treatments L45M and L45H in the experimental week they were switched, and for a good comparison this required synchronization.
It is evident that lambs in treatments L45M and L45H consumed more feed than animals on M and H, respectively, for several weeks after being switched from L, even though the BW in the groups were similar (Figure 2). The same figure also shows that the gains in BW in groups L45M and L45H exceeded those of lambs in groups M and H. This phenomenon is widely known as compensatory growth, which is well recorded although poorly understood (Graham and Searle, 1975; Moran and Holmes, 1978). Compensatory growth has been linked with greater feed efficiency (Kabbali et al., 1992a; Ryan et al., 1993a), sometimes in relation to a decrease in maintenance requirements (Ryan et al. 1993b), or an increase in mainly protein and water growth at the expense of lipid growth (Iason et al., 1992; Kabbali et al., 1992b; Ryan et al., 1993b). However, the observed compensatory growth in our sheep was associated with increased feed intake (Figure 2), as it has been observed by others (Marais et al., 1991; Iason et al., 1992; Ryan et al. 1993a). In addition, BCS increased very rapidly during the period of compensatory growth (Figure 2). Kyriazakis and Emmans (1992) suggested that animals, after a period of nutritional limitation, will try to achieve an intake that allows the restoration of the preferred lipid to lean ratio in their bodies. The greater intake of pellet-fed lambs that had previous received hay, combined with the rapid increase in BCS, would be consistent with such a view. Increased intake was associated with lambs in groups L45M and L45H initially being leaner than their respective controls when M and H were fed alone. Figure 2 shows that differences in BCS later decreased and that then ADFI became similar for lambs in L45M and M, and for lambs in L45H and H. In the case of L45M, BCS was greater than it was in M toward the end. At least part of this difference reflected the relatively low BCS of group M from the beginning of the experiment (Table 3).
Using Models to Describe the Effects of BW and BCS on the Intake of Hay and Pellets
It was our aim to break the correlation between BW and BCS that normally occurs when animals grow on a given feed. Figure 5a shows how the data we collected fit into the possible BW by BCS "space." For the evaluation of this graph it is important to consider that very high BW combined with low BCS, and vice versa, cannot be realized (i.e., part of the depicted BW by BCS space is inaccessible). Our success in breaking the relationship between BW and BCS was greater with L than with M and H.
View larger version (8K):
[in this window]
[in a new window]
|
Figure 5. Weekly smoothed BCS plotted against weekly BW by treatment group during the periods lambs consumed L (a), M (b), or H (c). L = chopped hay; M = oat-feed-based pellets; and H = barley-based pellets.
|
|
We aimed to develop a simple empirical model that can be used to describe the effects of animal fatness, as measured by BCS, on the feed intake by sheep that we observed. Our data on intake that were obtained suggested that the hay intake by sheep at a given BCS was directly proportional to BW (see Figure 3 and the previous discussion). It is not immediately evident how BCS affects intake by sheep of a given BW. However, hay intake data obtained with group H65L were suitable to suggest what the shape of this relationship could be. Mean BW of this treatment remained almost constant after lambs were switched to hay, whereas their BCS decreased (Figure 3). A plot of mean hay intake against BCS suggested that there was essentially a linear relationship between the 2 variables. A simple model that can accommodate these 2 observations is the first that we selected for our analyses: ADFI = a x BW x (1 – b x BCS).
From the literature, it is not clear whether an increase in fatness has proportionally different effects on the intake of feeds that differ in quality. As a first attempt, we therefore fitted the same model to the data obtained with the pelleted feeds and subsequently investigated whether Model 2 could improve the description of our findings. It is evident from Figure 5b and 5c that we were less able to produce variation in BCS at a given BW for feeds M and H, which makes the analyses of these data less strong.
Three versions of Model 1 were fitted to the combined data. A model with parameter a estimated per feed and parameter b across feeds described the data much better than the model with a estimated across feeds. This was to be expected because it is known that the intake by a sheep at a given BW and BCS is affected considerably by feed quality (ARC, 1980; NRC, 1987; Ketelaars and Tolkamp, 1992). Estimating b by feed gave a statistically significant improvement in the description of the data, but the estimates of b for the 3 feeds differed little (Table 4). Even though there was a large difference in quality between the hay and the pelleted feeds as judged from animal performance, the use of separate b’s for each feed led to only a very small increase in the total variation accounted for by the model and a very small reduction in the model RSD.
Regression of ADFI on BW and BCS, their squares, and their interactions with Model 2 showed that by increasing the number of model parameters from 3 (Model 2c) to 4, 6, or 8 resulted in only a trivial decrease in the RSD (Table 4). Model 2c gave a slightly better fit than Model 1 and indicated a significant effect of BCS alone (Table 4). The initial suggestion that there was no effect of feed on the manner in which BCS affected ADFI can, therefore, only be approximate at this stage. The great advantage of Model 1 is that only one parameter seems to be affected by feed composition, whereas all 3 in Model 2c might well be (Table 4). However, the variation in BCS at a given BW that was achieved for feeds M and H was too limited to really disentangle the separate effects that BW and BCS can have on ADFI of such feeds. Further data are required before a choice between Models 1 and 2c can be recommended.
Final Considerations
No other models that predict intake on the basis of BW and BCS are available and, therefore, a direct comparison with previous work is not possible. It proved interesting, however, to analyze how the description provided by our model compared with existing predictions based on BW alone that are currently in use for sheep, e.g., in the United Kingdom (ARC, 1980), United States (NRC, 1987), and Australia (CSIRO, 1990). In these predictive models, intake changes frequently, though not always, curvilinearly with BW. This could well result, at least in part, because these models incorporate any effect of changes in animal fatness on intake via the BW effect by the use of metabolic size or otherwise. Similarly, the use of BW alone in combination with BCS in our models may well have resulted from loading any curvilinearity in intake response to an increase in BW onto the BCS component. In view of Figure 5, the analysis of the data from the hay diet is much less likely to be affected by this problem than that from the 2 pelleted feeds. Further data that succeed in breaking the relationship between BW and BCS to a greater extent than we did are required to help resolve this issue.
In the present experiment, data were obtained with Greyface x Texel cross ewe lambs, and BW was used as an expression of animal size. It is evident that a given BW may represent different stages of growth in different genotypes, for instance with different mature sizes, and that this will likely affect intake. To account for such effects, the rules of genetic size scaling (Taylor, 1980; Emmans, 1997), or the concept of relative size as proposed by CSIRO (1990), can be used. In addition, the effect that BCS has on feed intake may well differ between male and female sheep, and this remains to be investigated.
We conclude that intake of hay as well as of pelleted feeds by a sheep with a given BW is affected considerably by the animal’s fatness. This means that BW alone is an insufficient description of the animal’s size that is relevant for feed intake predictions. The incorporation of sheep fatness, as estimated by BCS, in intake models should, therefore, improve the accuracy of intake predictions. This does not necessarily mean that BW and BCS are the only (or even the most important) characteristics to define the animal size that is most relevant for a given genotype; they are, however, the most readily available. This work demonstrates that these characteristics do have an effect on feed intake and show the importance of collecting information on animal fatness such a BCS to predict intake. As a first step, we used a simple 2-parameter model to describe the results we obtained with hay-fed animals. Although the variation in BCS at a given BW we obtained in lambs fed greater quality pelleted feeds was much more limited, the same model also gave a reasonable description of these data. Additional observations with other feed qualities and sheep genotypes are required, however, before general predictive models can be developed.
|
Footnotes
|
|---|
1 This research was carried out with the support of the Scottish Executive Environment and Rural Affairs Department, Edinburgh, UK. 
2 The authors are grateful to David Anderson, Terry McHale, and Lesley Deans for expert care of experimental animals and data collection.
3 Corresponding author: Bert.Tolkamp{at}sac.ac.uk
Received for publication September 13, 2005.
Accepted for publication February 6, 2006.
|
LITERATURE CITED
|
|---|
ARC. 1980. The Nutrient Requirements of Ruminant Livestock. CAB Int., Wallingford, UK.
Blaxter, K. L., V. R. Fowler, and J. C. Gill. 1982. A study of the growth of sheep to maturity. J. Agric. Sci. Camb. 98:405–420.
CSIRO. 1990. Ruminants. Feeding Standards for Australian Livestock. CSIRO Publ., Melbourne, Australia.
Emmans, G. C. 1997. A method to predict the food intake of domestic animals from birth to maturity as a function of time. J. Theor. Biol. 186:189–199.
Emmans, G. C., and I. Kyriazakis. 1995. The idea of optimisation of animals: Uses and dangers. Livest. Prod. Sci. 44:189–197.
Emmans, G. C., and I. Kyriazakis. 2001. Consequences of genetic change in farm animals on food intake and feeding behaviour. P. Nutr. Soc. 60:115–125.
Foot, J. Z. 1972. A note on the effect of body condition on the voluntary intake of dried grass wafers by Scottish Blackface ewes. Anim. Prod. 14:131–134.
Forbes, J. M. 1995. Voluntary Food Intake and Diet Selection in Farm Animals. CAB International, Wallingford, UK.
Graham, N. M., and T. W. Searle. 1975. Studies of weaner sheep during and after a period of weight stasis 1. Energy metabolism and nitrogen utilization. Aust. J. Agric. Res. 26:343–353.
Humphries, W. R., P. C. Morrice, and I. Bremner. 1988. A convenient method for the treatment of chronic copper poisoning in sheep using subcutaneous ammonium tetrathiomolybdate. Vet. Rec. 123:51–53.[Abstract]
Iason, G. R., A. R. Mantecon, J. A. Milne, D. A. Sim, A. D. M. Smith, and I. R. White. 1992. The effect of pattern of food supply on performance, compensatory growth and carcass composition of Beulah and Welsh mountain lambs. Anim. Prod. 54:235–241.
Kabbali, A., W. L. Johnson, D. W. Johnson, R. D. Goodrich, and C. E. Allen. 1992a. Effects of compensatory growth on some body component weights and on carcass and non-carcass composition of growing lambs. J. Anim. Sci. 70:2852–2858.[Abstract]
Kabbali, A., W. L. Johnson, D. W. Johnson, R. D. Goodrich, and C. E. Allen. 1992b. Effects of undernutrition and refeeding on weights of body parts and chemical components of growing Moroccan lambs. J. Anim. Sci. 70:2859–2865.[Abstract]
Ketelaars, J. J. M. H., and B. J. Tolkamp. 1992. Toward a new theory of feed intake regulation in ruminants. 1. Causes of differences in voluntary intake: Critique of current views. Livest. Prod. Sci. 30:269–296.[CrossRef]
Ketelaars, J. J. M. H., and B. J. Tolkamp. 1996. Oxygen efficiency and the control of energy flow in animals and man. J. Anim. Sci. 74:3036–3051.[Abstract]
Kyriazakis, I. 2003. What are ruminant herbivores trying to achieve through their feeding behaviour and food intake? Pages 153–173 in Proc. VIth Int. Symp. Nutr. Herbivores, Merida, Mexico. Universidad Autónoma de Yucatán, Mérida, México.
Kyriazakis, I., and G. C. Emmans. 1992. The growth of mammals following a period of nutritional limitation. J. Theor. Biol. 156:485–498.[CrossRef][Medline]
Lawes Agricultural Trust. 2004. GenStat Release 7.2. Harpenden, UK.
Lewis, R. M., G. C. Emmans, W. S. Dingwall, and G. Simm. 2002. A description of the growth of sheep and its genetic analysis. Anim. Sci. 74:51–62.
Lippi, E., C. H. Machado, and M. C. Araripe. 2003. Assessment of some clinical and laboratory variables for early diagnosis of cumulative copper poisoning in sheep. Vet. Hum. Toxicol. 45:289–293.[Medline]
Marais, P. G., H. J. Van der Merwe, and J. E. J. DuToit. 1991. The effect of compensatory growth on feed-intake, growth rate, body composition and efficiency of feed-utilization in Dorper sheep. S. Afr. J. Anim. Sci. 21:80–88.
Moran, J. B., and W. Holmes. 1978. The application of compensatory growth in grass/cereal production beef systems in the United Kingdom. World Rev. Anim. Prod. 14:65–73.
NRC. 1985. Nutrient Requirements of Sheep. Sixth rev. ed. Natl. Acad. Press, Washington, DC.
NRC. 1987. Predicting Feed Intake of Food-Producing Animals. Natl. Acad. Press, Washington, DC.
NRC. 2000. Nutrient Requirements of Beef Cattle. Update 2000 of the 7th rev. ed. 1996. Natl. Acad. Press, Washington, DC.
NRC. 2001. Nutrient Requirements of Dairy Cattle. 7th rev. ed. Natl. Acad. Press, Washington, DC.
Pittroff, W., and M. M. Kothmann. 2001. Quantitative prediction of feed intake in ruminants. 1. Conceptual and mathematical analysis of models for sheep. Livest. Prod. Sci. 71:131–150.
Russel, A. J. F., J. M. Doney, and R. G. Gunn. 1969. Subjective assessment of body fat in live sheep. J. Agric. Sci. (Camb.) 72:451–454.
Ryan, W. J., I. H. Williams, and R. J. Moir. 1993a. Compensatory growth in sheep and cattle. 1. Growth pattern and feed intake. Aust. J. Agric. Res. 44:1609–1621.
Ryan, W. J., I. H. Williams, and R. J. Moir. 1993b. Compensatory growth in sheep and cattle. 2. Changes in body composition and tissue weight. Aust. J. Agric. Res. 44:1623–1633.
Sibbald, A. M., and S. M. Rhind. 1997. The effect of previous body condition on appetite and associated insulin profiles in sheep. Anim. Sci. 64:247–252.
Taylor, S. C. S. 1980. Genetic size scaling rules in animal growth. Anim. Prod. 30:161–165.
Tolkamp, B. J., and J. J. M. H. Ketelaars. 1992. Toward a new theory of feed intake regulation in ruminants. 2. Costs and benefits of feed consumption: An optimization approach. Livest. Prod. Sci. 30:297–317.[CrossRef]
Velleman, P. F. 1980. Definition and comparison of robust non-linear data smoothing algorithms. J. Am. Stat. Assoc. 75:609–615.[CrossRef]
Vera, R. R., J. G. Morris, and L.-J. Koong. 1977. A quantitative model of energy intake and partition in grazing sheep in various physiological states. Anim. Prod. 25:133–153.[Medline]
Top
Abstract
INTRODUCTION
), H (
), H45M (
), and H65M (•). M = oat-feed-based pellets; and H = barley-based pellets. Treatments M and H received M or H, respectively, throughout. Treatments H45M and H65M switched from H to M when their BW reached 45 or 65 kg, respectively. The residual SD (RSD) is also given.
). L = chopped hay; M = oat-feed-based pellets; and H = barley-based pellets. Treatment L received L throughout. The other treatments switched from pellets to hay when their BW reached 95, 45, or 65 kg, respectively. The residual SD (RSD) is also given.