J. Anim. Sci. 2006. 84:767-776
© 2006 American Society of Animal Science
Identifying differences in feed efficiency among group-fed cattle
L. O. Tedeschi*,1,
D. G. Fox
,
M. J. Baker and
D. P. Kirschten
* Department of Animal Science, Texas A&M University, College Station 77843 and
and
Department of Animal Science, Cornell University, Ithaca, NY 14853
 |
Abstract
|
|---|
Identification of efficient animals in the postweaning growth phase for use in selection for improved feed efficiency is important to improve the economic and environmental sustainability of the beef cattle industry. Progeny testing using group-fed animals in commercial feedlots is the most common and practical method used to evaluate postweaning growth on large numbers of animals. We developed the Cornell Value Discovery System (CVDS) to dynamically predict growth rate, accumulated weight, days required to reach target body composition, carcass weight, and composition of individual beef cattle fed in group pens. Observed BW, ADG, BW at 28% empty body fat (EBF), breed type, environmental conditions, and dietary ME concentration are used by the CVDS to predict, for each animal in a pen, the feed DM required for maintenance (FFM), the feed DM required for gain, and the total DM required for maintenance and gain (DMR). The CVDS then computes DMR-to-ADG ratio (DMR:ADG), which is a feed conversion measure, and ADG-to-DMR ratio (ADG:DMR), which is a feed efficiency measure, for each animal. This study used the observed F:G ratio of 362 individually fed steers to evaluate CVDS-predicted indicators of feed efficiency and the Kleiber ratio. A subset of 37 data points was used to evaluate residual feed intake (RFI) as an indicator of feed efficiency. The database included 4 published studies, each with detailed individual animal description, environment, diet, and body composition information. The CVDS-predicted DMR:ADG accounted for 84% of the variation in the actual F:G ratio with a mean bias of 1.94% (P = 0.20). The predicted FFM to actual DMI ratio had a high correlation with actual ADG (R2 = 0.76), and indicated a decay-type nonlinear dilution of FFM as ADG increased. The CVDS-predicted ADG:DMR and the Kleiber ratio had a significant (R2 = 0.88) logarithmic relationship. In an analysis of a contemporary group within the database, RFI was highly correlated with the F:G ratio (r = 0.71). There was a positive relationship between RFI and EBF. The RFIM (DMI – DMR) was moderately correlated with DMI and ADG (0.37 and –0.38; respectively), suggesting that selecting for low RFIM would decrease DMI and increase ADG in this database. We conclude that the CVDS model can be used to identify differences in the F:G and G:F ratios by predicting DMR for individual growing cattle fed in groups.
Key Words: cattle • feed efficiency • growth • modeling • selection
|
INTRODUCTION
|
|---|
The conversion of feed into animal products during the postweaning growth phase has a large influence on the cost of producing beef (Tess and Kolstad, 2000
; Herd et al., 2003). The development of the USDA-EPA (http://cfpub2.epa.gov/npdes/afo/cafofinalrule.cfm) regulations for concentrated animal feeding operations to protect air and water quality has added pressure on the animal industries to reduce the nutrients fed for the same animal production. Methods are being developed to compute expected progeny difference for feed efficiency in cattle.
Residual feed intake (RFI) determined for individuals within a contemporary group with individual measurements of feed intake has been proposed as a method for selecting for efficient animals (Koch et al., 1963; Archer et al., 1997; Arthur et al., 2001a). Arthur et al. (2001a,b) reported that RFI is moderately heritable and genetically independent of growth traits. However, the cost of measuring feed intake in individual animals in commercial feedlots limits the use of including observed feed efficiency traits in breeding programs.
A growth model, the Cornell Value Discovery System (CVDS; Perry and Fox, 1997; Guiroy et al., 2001), was developed to predict feed required by individuals fed in feedlot group pens, based on observed growth, BW, and carcass measurements that can be readily obtained. This model has been successfully used to allocate feed costs to individual animals fed in pens (Fox et al., 2001, 2002). Recently, we published an enhanced, dynamic version of that model to improve its accuracy and usefulness in predicting differences in the F:G and G:F ratios among group-fed bulls, steers, and heifers (Tedeschi et al., 2004). The objective of this paper was to evaluate indicators of feed efficiency in young growing beef cattle predicted by the revised CVDS that contained our dynamic growth model and other methods that have been used to identify differences in feed efficiency.
|
MATERIALS AND METHODS
|
|---|
Four measures of feed efficiency during postweaning growth were evaluated and compared using data from 362 individually fed steers. These methods were feed efficiency indicators predicted by the CVDS, the observed feed-to-gain ratio (F:G, kg of DM/kg of ADG), the Kleiber ratio (KR), and RFI. In this paper, we assumed that feed conversion was F:G, and feed efficiency was the gain-to-feed ratio (G:F, g of ADG/kg of DM).
Evaluation of Feed Efficiency Indicators with the CVDS Growth Model
The dynamic, iterative, and mechanistic (DIM) growth model developed by Tedeschi et al. (2004) as implemented in the CVDS was used to compute indicators of feed efficiency for the 362 individually fed steers in the database. The DIM growth model was developed to dynamically predict daily growth rate and accumulated BW at any day on feed, days required to reach target body composition, carcass weight (CW), and composition of individual beef cattle for use in individual cattle management systems. The mathematical model can predict either ADG from DMI or DM required (DMR) for the observed ADG, BW, and empty body fat (EBF).
The size-scaling procedure of the NRC (2000) that computes energy requirements for growth is used in the CVDS; this procedure uses ADG, BW, and BW at 28% EBF to account for body size and energy content of gain in predicting the energy required for gain. Maintenance energy requirement was computed from BW, BCS, breed, physical activity, and environmental factors as described by Fox and Tylutki (1998). The ME concentration of the diet was used to compute dietary NEm and NEg, which were used to compute feed for maintenance (FFM) and feed for gain (FFG).
The CVDS uses the decay equation developed by Tedeschi et al. (2004) to compute a variable efficiency of ME to NEg for individual animals, based on the energy and protein content of gain. Indicators of feed efficiency such as the DMR-to-ADG ratio and the ADG-to-DMR ratio are computed from CVDS predictions of DMR. Both of these ratios account for differences in BW, energy content of gain, and partial efficiencies of ME use for protein and fat deposition.
Evaluation of Feed Efficiency with the Kleiber Ratio
This ratio of ADG to metabolic BW (Eq. [1]; Kleiber, 1936) was included in this evaluation because it does not require individual intake to be measured and has been used to identify animals with high efficiency of growth relative to body size. A high value indicates a greater dilution of maintenance energy requirement. This implies that as ADG increases at the same BW0.75, more growth is obtained without increased maintenance energy cost. In the current study, KR was calculated as:
 | [1] |
where BW0.75 is the mean metabolic BW (MMBW) during the growth period being evaluated.
Evaluation of Feed Efficiency with Residual Feed Intake
The RFI calculation (Eq. [2]) represents the difference between the feed consumption of an individual and the feed predicted to be required by that individual based on the multiple regression of observed DMI on ADG and MMBW of a group of cattle fed under the same conditions as initially defined by Koch et al. (1963; Eq. [3]). A multiple regression analysis of DMI on ADG and MMBW without the interaction (ADG x MMBW) component was used to obtain the RFI values using the data of Guiroy (2001; group 5 in Table 1; n = 37) because these animals were a contemporary group (same breed type that were fed the same diet and in the same location). In the current study, the RFI was calculated and the multiple regression of DMI on ADG and MMBW was performed as follows:
 | [2] |
and
 | [3] |
where RFI is residual feed intake (kg/d); DMI is measured dry matter intake (kg/d); 
is the predicted DMI (kg/d); ß0, ß1, and ß2 are the parameters of the multiple regression, and MBW0.75 is MMBW.
This approach was included in our evaluation because high genetic correlations between RFI and F:G have been reported (Arthur et al., 2001b), and it has been used in genetic evaluations (Kennedy et al., 1993; Veerkamp et al., 1995). Animals with a negative RFI consumed less feed than expected for their actual MMBW and ADG compared with their contemporaries; therefore, they were more efficient than their contemporaries. Efficiency traits were also compared with the modeled RFI (RFIM), which represents the difference between DMI and DMR.
Database Used to Evaluate Indicators of Feed Efficiency
The database consisted of 362 steers individually fed in 4 previously published studies (Nour et al., 1983; Perry et al., 1991; Perry and Fox, 1997; Guiroy et al., 2001). This database was utilized because it contained the information needed to evaluate predictions of feed efficiency by the CVDS (complete feed analysis, individual animal weights and intakes, environmental temperatures, and carcass chemical composition). This database is described in Table 1
.
Steers (240 animals) were individually fed in individual pens of 4.92 or 5.28 m2 in a slatted floor confinement barn or with electronic Calan Broadbent (American Calan, Northwood, NH) feeding gates (122 animals) outside in partially covered, paved group-pens of 278 m2. The CVDS-predicted NEm requirement was 6 to 12% higher for steers fed in the Calan-gate facility compared with the individual-pen facility (Tedeschi et al., 2004). All animals were allowed to consume their diets ad libitum. The dietary ME values (Guiroy et al., 2001; Table 1) were computed for each study using the Cornell Net Carbohydrate and Protein System (CNCPS) model (Fox et al., 2004), using the information available for feed composition.
Because of the intrinsic differences in conducting these studies, animals were assigned to 9 groups to better account for differences in behavior (individual vs. group), dietary ME (2.97, 2.96, 2.85, and 2.62 Mcal/kg), and housing type (inside vs. outside) as shown in Table 1. A standard reference weight of 478 kg was used for all steers. This is the weight of the standard reference animal used in the size-scaling procedure at small marbling (USDA low Choice quality grade), and is associated with 28% EBF as described in the NRC (2000). In this study, the BW at 28% EBF for each steer was assumed equal to their mature weight. Estimates of carcass and body composition were calculated based on chemical analysis of the 9th to 11th rib section of animals in the studies of Guiroy et al. (2001), Perry et al. (1991), and Perry and Fox (1997), whereas chemical analysis was performed on composite samples from complete grinding of one side of the carcass as in the study of Nour (1982).
The environmental information (hair depth, temperature, relative humidity, hair coat, sunlight, and wind) used in the CVDS model for the inside and outside housing locations is listed for each month in Table 2. The monthly average temperature for the period of growth of the animals housed in the outside location was obtained from the weather station of the Cornell University Teaching and Research Center (Harford, NY). To estimate the temperature for the animals housed inside the confinement barn for the same period of growth, the difference between outside and inside temperatures measured between 1999 and 2002 was applied to the temperature obtained from the weather station; thus, the temperature differential between outside and inside was kept constant across months. As shown in Table 2, the largest temperature difference (inside – outside) was between December and February, which include the growth periods of most of the animals in our evaluation database. A similar adjustment was made for relative humidity.
View this table:
[in this window]
[in a new window]
|
Table 2. Average hair depth (HD), temperature (T), and relative humidity (RH) used by the growth model for animals located inside and outside the experimental barns1
|
|
Statistical Analyses and Model Evaluation
All statistical analyses were performed with SAS (SAS Inst., Inc., Cary, NC) and regression parameters were estimated by PROC REG. The regression through the origin was obtained using the NOINT option in PROC REG (SAS Inst. Inc.). The statistical comparison between observed and predicted values was performed using the 2-sample t-test assuming different variances (Neter et al., 1996). Model adequacy, as discussed and implemented by Tedeschi (2006), was assessed with linear regression statistics (Harrison, 1990; Mayer and Butler, 1993; Mayer et al., 1994), the concordance correlation coefficient (Lin, 1989; Liao, 2003), and the mean square error of prediction (Theil, 1961). Additionally, ranking analysis was performed to compare model-predicted and observed values using nonparametric approaches. In this analysis, Spearman’s rs statistic and Kendall’s 
statistic were used to assess the correlation of the ranked values, and the Kolmogoroff-Smirnov test was used to verify the distribution similarity between predicted- and observed-ranked values (Agresti, 1996, 2002), as implemented by Tedeschi (2006).
|
RESULTS
|
|---|
Feed Efficiency Indicators Evaluated with the CVDS Growth Model
Table 3 provides the Pearson correlation coefficients for feed efficiency indicators evaluated. Figure 1 compares observed F:G and predicted DMR:ADG by the DIM growth model for the 362 individually fed steers. The DIM growth model accounted for 84% (r2 = 0.84) of the variation in F:G with a mean bias of 1.94% (P = 0.19; Figure 1A). The accuracy and concordance between predicted DMR:ADG and observed F:G were high (Cb = 0.99 and rc = 0.91; respectively). Figure 1B shows the ranking of observed F:G and predicted DMR:ADG values for the 362 individually fed steers. This r2 was also high (0.81), showing a good agreement between the ranks of predicted and observed values. The Spearman’s rS of 0.90 (P < 0.001) and Kendall’s of 0.73 (P < 0.001) indicated the ranks of observed F:G and predicted DMR:ADG were highly correlated and had similar distribution (Kolmogoroff-Smirnov value = 0.08; P = 0.11). The DIM growth model was able to correctly identify 34 of the best 50 animals, implying that 68% of the ranks of predicted DMR:ADG and observed F:G were within the 50 most efficient animals. These results were similar for observed G:F and predicted ADG:DMR ranking analysis; therefore, the data are not shown.
A simulation with empty BW gain, which varied from 0.8 to 2.2 kg/d, indicated a high correlation between FFG and empty BW gain (r2 = 0.99), which is accounted for in the CVDS. This was expected because the predicted FFG divided by ADG will increase as ADG increases, due to increased fat content of the gain. The ratio of predicted FFM to actual DMI had a high correlation with actual ADG (R2 = 0.76), and indicated a nonlinear dilution of FFM compared with DMI (decay form) as ADG increases (Eq. [4]). A similar decay shape was observed for the ratio of predicted FFM to DMR with actual ADG (R2 = 0.90; not shown):
 | [3] |
Figure 2 shows the relationship between observed F:G and model-predicted DMR:ADG for animals in the database that were at the same stage of growth (fed from 60 to 100% of mature BW). Only a small proportion of the steers in the database was fed over this same stage of growth. In Figure 2A, a variability of ± 10% in the degree of maturity was assumed, whereas in Figure 2B, it was more stringent at ± 5% of variation. Overall, there was a high agreement (precision) between the observed F:G and the model-predicted DMR:ADG (r2 = 0.74 and 0.85, respectively, for the 10 and 5% levels of variability in the degree of maturity). Despite the model underprediction, accuracy was also high (Ar = 0.85 and 0.80, respectively for the 10 and 5% levels of variability in the degree of maturity). The rank analysis indicated the growth model was able to identify correctly 70 and 100% of the one-third most efficient animals at a variability of 5 and 10%, respectively, from being fed from 60 to 100% of their mature BW during the test.
View larger version (15K):
[in this window]
[in a new window]
|
Figure 2. Relationship between observed feed-to-gain ratio (F:G, kg/kg) and the ratio of model-predicted DM required to ADG (DMR:ADG, kg/kg) for animals at the same stage of growth assuming a variability of A) ± 10% or B) ±5% for the thresholds for degree of maturity (0.6 and 1.0). The solid line indicates Y = X (the unity line).
|
|
Feed Efficiency Evaluated with the Kleiber Ratio
The relationship between observed G:F and KR was high (R2 = 0.80; plot not shown). Figure 3A shows the relationship between model-predicted ADG:DMR and KR because ADG is used as the numerator in both calculations. The CVDS-predicted ADG:DMR and KR had a significant (R2 = 0.88) logarithmic relationship. Figure 3B shows a decay relationship between DMR:ADG and KR with an R2 = 0.80. These relationships suggest that ADG:DMR is better correlated with KR than is DMR:ADG. The rank analysis (Figure 3C) indicated high correlation between predicted ADG:DMR and KR with rS of 0.93 (P < 0.001) and of 0.79 (P < 0.001). An increase in the KR reflects a greater ADG relative to maintenance requirement, because the denominator is MMBW. A similar result was obtained for DMR:ADG and KR relationship.
Figure 4 shows the relationship between observed G:F and model-predicted ADG to FFM ratio (ADG:FFM). As with the KR, the ADG relative to FFM increases more rapidly than actual ADG relative to DMR, because it does not account for the energy required for growth. Using a nonlinear function, a plateau of 217 g/kg was obtained for this relationship in this database (Figure 4).
Feed Efficiency Evaluated with Residual Feed Intake
The regression of DMI on ADG and MMBW (n = 37; R2 = 0.45; and MSE = 0.432) that was used to compute RFI for each animal in the database is shown in Eq. [5]:
 | [5] |
where is the multiple-regression predicted DMI (kg/d), and MBW0.75 is the mean metabolic BW (kg0.75). The intercept and the coefficient of MMBW were not significantly different from zero (P = 0.38 and 0.22, respectively).
Figure 5 indicates that RFI accounted for 51% of the variation in observed F:G and only 11% of the variation in model-predicted DMR:ADG in this database. Inclusion of EBF (Eq. [6]) was highly significant (P = 0.003), indicating that DMI increased with increases in EBF, which is logical because of the increased energy requirement for gain associated with it. The r2 (0.57) increased substantially when EBF was included in the equation:
View larger version (14K):
[in this window]
[in a new window]
|
Figure 5. Relationship of residual feed intake (RFI, kg/d) and A) observed and B) model-predicted feed-to-gain ratio (F:G, kg/kg) using data from Guiroy (2001).
|
|
 | [6] |
where is multiple-regression predicted dry matter intake (kg/d); EBF is empty body fat (%); and MBW0.75 is the mean metabolic BW (kg0.75). The intercept and MMBW were not significantly different from zero (P = 0.11 and 0.97; respectively); all other coefficients were highly different from zero.
Unlike the RFI computed with the use of Eq. [5], RFI computed with EBF (Eq. [6]) explained less (r2 = 40%) of the variation in observed F:G.
|
DISCUSSION
|
|---|
In this study, DMR:ADG predicted by the CVDS growth model accounted for 84% of the variation in observed F:G. We believe this occurred because with this data set the model accounted for most of the variables that affected their energy requirements (e.g., BW, breed, and environmental effects on maintenance requirement, effect of proportion of BW at 28% EBF on energy content of gain, and efficiency of use of ME for fat and protein deposition). Williams et al. (2005) reported that DMR predicted by the Decision Evaluator for the Cattle Industry and CVDS models accounted for 53 and 44% of the variation in the observed DMI, with high genetic correlations (0.96 and 0.95, respectively).
The steers in this database that had the genetic capability to grow more rapidly at the same stage of growth consumed more feed to meet their increased demand for nutrients for growth, so DMI increased proportionally to meet this demand. Therefore, feed efficiency improved because of increased weight gain relative to their maintenance requirement. The high proportion of variation in observed F:G that was accounted for with little bias by predicted DMR:ADG in this study indicated that the animals with higher growth potential had the digestive and metabolic capacity to utilize more nutrients with little change in digestive and metabolic efficiency. The equations used to predict partial efficiencies of use of DE and ME for NEm and NEg were developed over 30 yr ago in serial slaughter and metabolism studies at the University of California (Garrett, 1980). These relationships are used by the NRC (2000, 2001) and the CVDS (Tedeschi et al., 2004), and accurately predict growth rate and feed requirements for the current beef cattle population, which has higher body size and growth rate than the animals used to develop those relationships (NRC, 2000). This agrees with observations with the current population of dairy cows. The partial efficiencies of use of DE and ME for NEl developed over 30 yr ago in the USDA Beltsville respiration chambers (Moe, 1981) accurately predict milk production of current dairy cows that have increased in size and have been selected for higher milk production (Fox et al., 2004).
Veerkamp and Emmans (1995) concluded that there is no strong evidence for assuming that partial efficiency of energy use for different physiological functions (e.g., growth, lactation, reserves, maintenance) is genetically different either inter- or intraanimal species, but rather it is likely that most of the genetic variation between animals is due to differences in milk yield, capacity of feed intake, extent of tissue mobilization, and differences in partitioning between these components. They further concluded that most of the high genetic merit animals are more energetically efficient because they partition the available energy better than do low genetic merit animals. Differences in energy partition to heat, protein, or fat deposition is strongly associated with maturity degree (Webster, 1980). Ferrell (2003) concluded that progress in reduction of energy losses in ruminants would likely require incremental changes in systems (animal management, feeding strategies, and diet characteristics), rather than in major genes. Although variation exists in various components of animal digestion and metabolism, these processes are highly regulated, integrated systems. There is some evidence for differences in digestibility of feed due to breed. Several reports (Ashton, 1962; Karue et al., 1972; Frisch and Vercoe, 1977; Moran et al., 1979) have indicated that Bos indicus utilized low quality forage diets more efficiently than did Bos taurus but the difference is reduced when high quality forage is fed (Moore et al., 1975; Krehbiel et al., 2000). This is likely to occur because important differences exist in energy and nutrient availability depending on site of digestion and absorption. Hegarty (2001) indicated that there are genotype differences regarding the mean retention time, which is an important factor affecting digestibility of the diet, consequently altering productivity and selection traits. Nonetheless, there is evidence of genetic variation in F:G and RFI (Arthur et al., 2001a,b; Schenkel et al., 2004) with heritability varying from low to moderate.
The KR had a high relationship with CVDS-predicted ADG:DMR and DMR:ADG ratios. The KR accounts for dilution of maintenance requirements by increased ADG. However, it does not account for differences in fat and protein content of gain. The result of using this type of approach has resulted in selection for large-frame animals, without necessarily changing growth efficiency (Webster et al., 1982). This is because larger, faster-gaining animals are likely to be leaner at a particular weight and require less feed per kilogram of gain, which would appear to increase their feed efficiency. There has been little difference in efficiency of gain among animals of different sizes when they were fed to the same body composition (Klosterman, 1972).
In this study, the correlation between RFI and F:G (r = 0.72) was comparable to other reported values. Basarab et al. (2003) indicated that RFI is moderately and positively correlated with F:G (range of r = 0.42 to 0.70). They also reported that RFI was correlated with DMI (range of r = 0.42 to 0.72); in our study, this correlation was 0.74 (Table 3). However, the correlation between DMI and RFI does not explain the biological reasons for variability in RFI. It represents the residual in a multiple regression analysis and contains all of the variation not accounted for (e.g., energy content of gain, differences in efficiency of use of ME for fat and protein in the gain, depression in feed digestibility, compensatory gain, protein turnover) by the variables used in the regression equation. Therefore, it should not be assumed that a negative RFI reflects a lower than expected maintenance requirement. Nonetheless, it is possible that genetic variation in animal physical activity, heat production associated with proportional differences in visceral organ mass (e.g., compensatory growth), and cellular metabolism (e.g., mitochondrial proton leakage) are associated with efficient animals and that RFI accounts for some of this variation.
There is some indication that lower RFI steers are leaner than higher RFI steers, indicating an association of genetic selection for RFI and maturity patterns (Richardson et al., 1998). Our analysis supports this finding because the coefficient for EBF was significant and the Pearson correlation between RFI and EBF was significantly different from zero (r = 0.42; Table 3), indicating that, on average, DMI would be higher for fatter animals at the same BW and ADG; this would be expected because of a higher energy requirement for gain. The correlation between RFI and EBF was low, which is in agreement with literature data, suggesting that the magnitude of variation in RFI attributable to variation in body composition of growth is small (5%; Herd et al., 2004).
Richardson et al. (1998) and Basarab et al. (2003) reported that differences of approximately 25% in intake still existed even when BW, ADG, and body compositional differences are accounted for. Basarab et al. (2003) reported that RFI was positively related with DMI, MEI, HP, and RE, which indicates that the higher the DMI, the higher is the HP and energy retained. The magnitude of the correlation was less between RFI and RE (r = 0.28) compared with the correlation between RFI and HP (r = 0.56) and MEI (r = 0.80) when RFI was not adjusted for variation in body composition. When RFI was adjusted for variation in body composition, RFI was strongly correlated with MEI and HP (r = 0.70 for both), but not with RE (r = 0.02; Basarab et al., 2003). Therefore, genetic selection for animals based on a negative RFI without adjusting for body composition and considering an acceptable performance may lead to selecting animals with lower DMI and possibly production, because the ability to retain energy is highly correlated with heat production.
Studies are underway to improve the RFI approach by developing regression components that can account for more of the variables that affect energy requirements, such as body fat and IGF-I. Our principal concern about relying on RFI alone to select for genetic merit is the need to know actual DMI, which limits the number of animals that can be evaluated because in most feedlot conditions it is too difficult and expensive to measure DMI of individuals.
The Pearson correlation coefficients of RFIM (Table 3) with DMI, ADG, and F:G were significantly greater than zero (0.37, –0.38, and 0.86; respectively, P < 0.01) and very similar to those values reported by Arthur et al. (2001b). These coefficients suggest that selecting for low RFIM would decrease DMI and increase ADG. The correlation coefficient between RFI and RFIM was high, but RFIM was not correlated with EBF (P = 0.70), suggesting that body composition would not change when selecting for RFIM.
We conclude that the use of biologically based nutrition models that account for the primary variables that influence energy requirements can be useful in identifying differences between animals in the F:G and G:F ratios. Additional enhancements are needed to increase the accuracy of our growth model to account for more of the unexplained variability in observed feed efficiency. Included are 1) more accurate estimation of initial and final EBF as a function of mature weight, frame size, and gain; 2) development of better equations to adjust ADG requirements of animals in a contemporary group to the same stage of growth; and 3) development of mechanistic equations to more accurately estimate differences in maintenance requirements of individual animals. Except for breed effects, differences in basal maintenance requirements between group-fed animals cannot be predicted with the current structure of our growth model.
|
IMPLICATIONS
|
|---|
A growth model that accounts for the principal sources of variation in energy requirements can be used to accurately identify differences in feed-to-gain and gain-to-feed ratios between individually growing cattle fed in groups. In this study, the Cornell Value Discovery System-predicted dry matter required to average daily gain ratio accounted for 84% of the variation in actual feed conversion (feed:gain) with a mean bias of 1.94%. Feed-to-gain or gain-to-feed ratio comparisons should be made over the same stage of growth to avoid bias due to variation in proportion of fat in the gain during the test period. Further improvements can be made in our model to account for more of the variation in observed feed-to-gain and gain-to-feed ratios, requiring data from experiments with individually fed animals in which observed dry matter intake, accurate feed characterization, composition of growth, and other animal information are available.
1 Corresponding author: luis.tedeschi{at}tamu.edu
Received for publication June 29, 2004.
Accepted for publication October 24, 2005.
|
LITERATURE CITED
|
|---|
Agresti, A. 1996. An introduction to categorical data analysis. Wiley-Interscience, New York, NY.
Agresti, A. 2002. Categorical data analysis. 2nd ed. John Wiley & Sons, New York, NY.
Archer, J. A., P. F. Arthur, R. M. Herd, P. F. Parnell, and W. S. Pitchford. 1997. Optimum postweaning test for measurement of growth rate, feed intake, and feed efficiency in British breed cattle. J. Anim. Sci. 75:2024–2032.[Abstract/Free Full Text]
Arthur, P. F., J. A. Archer, D. J. Johnson, R. M. Herd, E. C. Richardson, and P. F. Parnell. 2001a. Genetic and phenotypic variance and covariance components for feed intake, feed efficiency, and other postweaning traits in Angus cattle. J. Anim. Sci. 79:2805–2811.[Abstract/Free Full Text]
Arthur, P. F., G. Renand, and D. Krauss. 2001b. Genetic and phenotypic relationships among different measures of growth and feed efficiency in young Charolais bulls. Livest. Prod. Sci. 68:131–139.
Ashton, G. C. 1962. Comparative nitrogen digestibility in Brahman, Brahman x Shorthorn, Africander x Hereford and Hereford steers. J. Agric. Sci. 58:333–341.
Basarab, J. A., M. A. Price, J. L. Aalhus, E. K. Okine, W. M. Snelling, and K. L. Lyle. 2003. Residual feed intake and body composition in young growing cattle. Can. J. Anim. Sci. 83:189–204.
Ferrell, C. L. 2003. Energy partitioning in ruminants as related to feed intake. Paper presented at Symp. Factors Affecting Beef Efficiency in Beef Cattle, Can. Soc. Anim. Sci. Annu. Mtg. Jul. 10–13, 2003. Can. Soc. Anim. Sci. Saskatoon, Canada.
Fox, D. G., L. O. Tedeschi, and M. J. Baker. 2002. Determining post-weaning efficiency of beef cattle. Pages 44–66 in Beef Improvement Federation, 34th Annu. Mtg., Jul. 10–13, 2002. Beef Improv. Fed., Omaha, NE.
Fox, D. G., L. O. Tedeschi, and P. J. Guiroy. 2001. Determining feed intake and feed efficiency of individual cattle fed in groups. Pages 80–98 in Beef Improvement Federation Annu. Mtg. Jul. 11–14, 2001, San Antonio, TX. Beef Improv. Fed., San Antonio, TX.
Fox, D. G., L. O. Tedeschi, T. P. Tylutki, J. B. Russell, M. E. Van Amburgh, L. E. Chase, A. N. Pell, and T. R. Overton. 2004. The Cornell Net Carbohydrate and Protein System model for evaluating herd nutrition and nutrient excretion. Anim. Feed Sci. Technol. 112:29–78.
Fox, D. G., and T. P. Tylutki. 1998. Accounting for the effects of environment on the nutrient requirements of dairy cattle. J. Dairy Sci. 81:3085–3095.[Abstract]
Frisch, J. E., and J. E. Vercoe. 1977. Food intake, eating rate, weight gains, metabolic rate and efficiency of feed utilization in Bos taurus and Bos indicus crossbred cattle. Anim. Prod. 25:343–358.
Garrett, W. N. 1980. Energy utilization by growing cattle as determined in 72 comparative slaughter experiments. Pages 3–7 in Proc. Energy Metabolism, 8, Cambridge. EAAP Publ. No. 26, Butterworths & Co., London, UK.
Guiroy, P. J. 2001. A system to improve local beef production efficiency and consistency in beef quality and its implementation through the creation of a strategic alliance. Ph.D. Dissertation, Cornell University, Ithaca, NY.
Guiroy, P. J., D. G. Fox, L. O. Tedeschi, M. J. Baker, and M. D. Cravey. 2001. Predicting individual feed requirements of cattle fed in groups. J. Anim. Sci. 79:1983–1995.[Abstract/Free Full Text]
Harrison, S. R. 1990. Regression of a model on real-system output: An invalid test of model validity. Agric. Syst. 34:183–190.
Hegarty, R. S. 2001. Genetic diversity in function and microbial metabolism of the rumen. Pages 63–68 in Feed Efficiency in Beef Cattle; Proc. Feed Efficiency Workshop, Armidale, Australia. CRC for Cattle and Beef Quality, C. J. Hawkins Homestead, University of New England, Armidale, Australia.
Herd, R. M., J. A. Archer, and P. F. Arthur. 2003. Reducing the cost of beef production through genetic improvement in residual feed intake: Opportunity and challenges to application. J. Anim. Sci. 81(E Suppl.):E9–E17.[Abstract/Free Full Text]
Herd, R. M., V. H. Oddy, and E. C. Richardson. 2004. Biological basis for variation in residual feed intake in beef cattle. 1. Review of potential mechanisms. Aust. J. Exp. Agric. 44:423–430.
Karue, C. N., J. L. Evans, and A. D. Tillman. 1972. Metabolism of nitrogen in Boran and in Hereford-Boran crossbred steers. J. Anim. Sci. 35:1025–1030.
Kennedy, B. W., J. H. L. Van der Werf, and T. H. E. Meuwissen. 1993. Genetic and statistical properties of residual feed intake. J. Anim. Sci. 71:3239–3250.[Abstract]
Kleiber, M. 1936. Problems involved in breeding for efficiency of food production. Pages 247–258 in Proc. Am. Soc. Anim. Prod., Madison, WI.
Klosterman, E. W. 1972. Beef cattle size for maximum efficiency. J. Anim. Sci. 34:875–880.[Abstract/Free Full Text]
Koch, R. M., L. A. Swinger, D. Chambers, and K. E. Gregory. 1963. Efficiency of feed use in beef cattle. J. Anim. Sci. 22:486–494.[Abstract/Free Full Text]
Krehbiel, C. R., K. K. Kreikemeier, and C. L. Ferrell. 2000. Influence of Bos indicus crossbreeding and cattle age on apparent utilization of high-grain diet. J. Anim. Sci. 78:1641–1647.[Abstract/Free Full Text]
Liao, J. J. Z. 2003. An improved concordance correlation coefficient. Pharmaceut. Stat. 2:253–261.
Lin, L. I.-K. 1989. A concordance correlation coefficient to evaluate reproducibility. Biometrics 45:255–268.[Medline]
Mayer, D. G., and D. G. Butler. 1993. Statistical validation. Ecol. Modell. 68:21–32.
Mayer, D. G., M. A. Stuart, and A. J. Swain. 1994. Regression of real-world data on model output: An appropriate overall test of validity. Agric. Syst. 45:93–104.
Moe, P. W. 1981. Energy metabolism of dairy cattle. J. Dairy Sci. 64:1120–1139.
Moore, R. L., H. W. Essig, and J. L. Smithson. 1975. Influence of breeds of cattle on ration utilization. J. Anim. Sci. 41:203–207.[Abstract/Free Full Text]
Moran, J. B., B. W. Norton, and J. V. Nolan. 1979. The intake, digestibility and utilization of a low quality roughage by Brahman cross, Buffalo, Bonteng, and Shorthorn steers. Aust. J. Agric. Res. 30:333–340.
Neter, J., M. H. Kutner, C. J. Nachtsheim, and W. Wasserman. 1996. Applied Linear Statistical Models. 4th ed. McGraw-Hill Publishing Co., Boston, MA.
Nour, A. Y. M. 1982. Bovine carcass characteristics, quality, yield, palatability, chemical composition, and mineral profile of early and late maturing breeds of cattle fed two diets in two locations and serially slaughtered over a wide weight range. Ph.D. Dissertation, Cornell University, Ithaca, NY.
Nour, A. Y. M., and M. L. Thonney. 1987. Carcass soft tissue and bone composition of early and late maturing steers fed two diets in two housing types and serially slaughtered over a wide weight range. J. Agric. Sci. 109:345–356.
Nour, A. Y. M., M. L. Thonney, J. R. Stouffer, and W. R. C. White, Jr. 1983. Changes in carcass weight and characteristics with increasing weight of large and small cattle. J. Anim. Sci. 57:1154–1165.[Abstract/Free Full Text]
NRC. 2000. Nutrient Requirements of Beef Cattle. Updated 7th ed. National Academy Press, Washington, DC.
NRC. 2001. Nutrient Requirements of Dairy Cattle. 7th ed. National Academy Press, Washington, DC.
Perry, T. C., and D. G. Fox. 1997. Predicting carcass composition and individual feed requirement in live cattle widely varying in body size. J. Anim. Sci. 75:300–307.[Abstract/Free Full Text]
Perry, T. C., D. G. Fox, and D. H. Beermann. 1991. Effect of an implant of trenbolone acetate and estradiol on growth, feed efficiency, and carcass composition of Holstein and beef steers. J. Anim. Sci. 69:4696–4702.[Abstract]
Richardson, E. C., R. M. Herd, J. A. Archer, R. T. Woodgate, and P. F. Arthur. 1998. Steers bred for improved net feed efficiency eat less for the same feedlot performance. Anim. Prod. Austral. 22:213–216.
Schenkel, F. S., S. P. Miller, and J. W. Wilton. 2004. Genetic parameters and breed differences for feed efficiency, growth, and body composition traits of young beef bulls. Can. J. Anim. Sci. 84:177–185.
Tedeschi, L. O. 2006. Assessment of the adequacy of mathematical models. Agric. Syst. (In press).
Tedeschi, L. O., D. G. Fox, and P. J. Guiroy. 2004. A decision support system to improve individual cattle management. 1. A mechanistic, dynamic model for animal growth. Agric. Syst. 79:171–204.
Tess, M. W., and B. W. Kolstad. 2000. Simulation of cow-calf production systems in a range environment: II. Model evaluation. J. Anim. Sci. 78:1170–1180.[Abstract/Free Full Text]
Theil, H. 1961. Economic forecasts and policy. Pages 6–48 in Contributions to Economic Analysis. 2nd ed. R. Strotz, J. Tinbergen, P. J. Verdoorn, and H. J. Witteveen, ed. North-Holland Publishing Company, Amsterdam, The Netherlands.
Veerkamp, R. F., and G. C. Emmans. 1995. Sources of genetic variation in energetic efficiency of dairy cows. Livest. Prod. Sci. 44:87–97.
Veerkamp, R. F., G. C. Emmans, A. R. Cromie, and G. Simm. 1995. Variance components for residual feed intake in dairy cows. Livest. Prod. Sci. 41:111–120.
Webster, A. J. F. 1980. The energetic efficiency of growth. Livest. Prod. Sci. 7:243–252.
Webster, A. J. F., A. A. M. Ahmed, and J. P. Frappell. 1982. A note of growth rates and maturation rates in beef bulls. Anim. Prod. 35:281–286.
Williams, C. B., G. L. Bennett, T. G. Jenkins, L. V. Cundiff, and C. L. Ferrell. 2005. Using simulation models to predict feed intake: Phenotypic and genetic relationships between observed and predicted values. J. Anim. Sci. 83(Suppl. 1):13. (Abstr.)[Abstract/Free Full Text]
This article has been cited by other articles:
|
|
|
|
 
M. J. Baker, L. O. Tedeschi, D. G. Fox, W. R. Henning, and D. J. Ketchen
Using ultrasound measurements to predict body composition of yearling bulls
J Anim Sci,
October 1, 2006;
84(10):
2666 - 2672.
[Abstract]
[Full Text]
[PDF]
|
|
|
Top
Abstract
INTRODUCTION
. The solid line indicates Y = X (the unity line).
