| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
ANIMAL GENETICS |
* Laboratory of Animal Breeding and Genetics, Graduate School of Agricultural Science, Tohoku University, Aoba-ku, Sendai 981-8555, Japan;
and
Miyagi Prefecture Animal Industry Experiment Station, Tamatsukuri-gun, Miyagi 989-6445, Japan; and
Graduate School of Natural Science and Technology, Okayama University, 700-8530, Japan
 |
Abstract |
|---|
 Top Abstract INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Key Words: Duroc pig • efficiency of gain • genetic parameter • growth
|
INTRODUCTION |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
). It is necessary, therefore, to consider avenues for reducing feed input to improve production efficiency.
Estimates of genetic parameters for important traits are requisite for accelerating the progress of ongoing breeding programs. Selection for growth rate has received considerable attention in most pig improvement projects. For growth traits, numerous heritability and correlation estimates are available in pigs (Suzuki et al., 2005). However, there is a lack of reports on estimates of the genetic relationships among measures of the efficiency of gain and their component traits in pigs.
Selection for residual feed intake (RFI), that is, feed consumed above or below the requirements for production and maintenance, may be a useful alternative measure (Arthur et al., 2001b). This difference in intake is generally calculated as RFI by comparing the actual and predicted feed intake of an animal. The predicted feed intake is the amount of feed that the animal is expected to consume by its BW and growth rate. Variation in RFI may reflect differences in the efficiency with which animals digest and utilize energy for maintenance and production. Selection against RFI might improve the efficiency of energy utilization without reducing appetite, which is needed for production of the product. Inclusion of this trait in a breeding program depends on its heritability and its relationships with other traits of interest.
The objective of this study was to estimate genetic parameters for and genetic relationships between measures of the efficiency of gain and its component traits in Duroc pigs.
|
MATERIALS AND METHODS |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Animals at Performance Test
Duroc pigs used in this study were from a line that was selected through 7 generations, from 1995 to 2001, at the Miyagi Prefecture Animal Industry Experiment Station. The average population size of each generation was 15.6 sires and 44.5 dams. A total of 1,642 pigs were tested, and the tested pigs were the progeny of 125 boars and 356 gilts. To estimate the variance components, a pedigree file was constructed. The total number of pigs in the pedigree, including test pigs, was 1,780. Description of the data structure is presented in Table 1. Gilts farrowed only once, and boars were retained for use during one 4- to 6-wk breeding period. Thereby, a new generation was obtained each year.
|
Selection Method
The detailed procedure for selection has been described by Suzuki et al. (2005). The selection was conducted without a control line. The first and second generations of selection were performed using an index selection method based on the relative desired gains (Yamada et al., 1975). Traits as selection criteria were ADG from 30 to 105 kg of BW, LM area (EMA), backfat thickness (BF) at 105 kg of BW, and i.m. fat (IMF). Genetic and phenotypic parameters used to derive the selection criteria were obtained from the performance test data of the first and second generations. The means of ADG, EMA, BF, and IMF at the first generation were 850 g, 36.1 cm2, 2.34 cm, and 4.3%, respectively. Relative desired gain was 135 g, 3.9 cm2, –0.54 cm, and 0.7%, for ADG, EMA, BF, and IMF, respectively, established from the performance test data of the first generation. Consequently, the selection index equation was as follows:
 |
Breeding values of ADG, EMA, BF, and IMF were estimated using multiple-trait, animal-model BLUP from the third generation onward. The breeding values were calculated using the prediction and estimation program after estimating genetic parameters using the variance component estimation (VCE) program (version 4.2.5; Neumaier and Groeneveld, 1998), with the models of generation and sex as fixed effects and random effects of the individual additive genetic effect and error. Relative economic weights of selection traits were calculated from the relative desired gains of ADG, EMA, BF, and IMF, which were established from performance test data of the first generation, as described above. The aggregate breeding values were calculated by multiplying the relative economic weights by the EBV of each trait; then, selection was executed. To avoid rapid disappearance of the genes of the base generation from the population, selection was made within sires for boars and within litters for gilts at the first generation. Approximately 15 boars and 50 gilts were selected at each generation. In each generation, inbreeding coefficients for individual pigs were computed. Based on inbreeding information, all matings were planned to minimize the rate of increase in inbreeding.
Traits in Study
The weekly BW of the individual pig during the test period was recorded and the ADG for each animal was calculated individually from the difference between beginning and end test BW divided by the number of days. The BW at midtest (MBW) for each animal was computed as the average of the beginning and end of test BW. Metabolic BW at midtest (MWT) was calculated as MBW raised to the power of 0.75 (i.e., MBW0.75). Daily feed intake (FI) was measured only for individual boars as the difference between supplied and leftover feed. The FI was then combined with ADG and MWT to compute measures of the efficiency of gain:feed conversion ratio (FCR) and RFI for individual boars. The FCR was calculated as FI divided by ADG. The RFI was estimated as nutritional RFI (RFInut), phenotypic RFI (RFIphe), and genetic RFI (RFIgen). The RFInut was estimated using the following method from JFSS (2005):
 |
where DE = the daily energy requirement = (140 x MWT) + (PR/0.42) + (FR/0.71); PR = protein accumulation = (–0.121 x MBW) + [119.2 x (ADG + 25.5)] x 5.66; FR = fat accumulation = (–0.268 x MBW) – (0.0015 x MBW2) + (99.65 x ADG) + (42.43 x ADG2) + [3.45 x (ADG – 21.4)] x 9.46; and 1 kg of feed contained 3,495 kcal of GE).
The RFIphe and RFIgen were estimated by the residual of phenotypic and genetic regression, respectively, from the multivariate analyses of FI on MWT and ADG, fitting a model with fixed terms for the overall mean plus random terms for the generation and the animal effects (including pedigree). Then, RFIphe was estimated as:
 |
where the phenotypic regression coefficients, 
; P = the phenotypic covariance matrix of 2 production traits (MWT and ADG) and c = the vector of the phenotypic covariance of FI with production traits. RFIgen was estimated as:
 |
where the genetic regression coefficients, 
; G = the genetic covariance matrix of MWT and ADG and c = the vector of the genetic covariance of FI with production traits estimated using the REML method.
Statistical Analysis
The covariance components for all traits were estimated by the REML method with the VCE program (Neumaier and Groeneveld, 1998). Standard errors of the heritability estimates and genetic correlations were also estimated using the same program. The (co)variance components for the efficiency of gain traits (RFInut, RFIphe, RFIgen, and FCR) were estimated using a 4-trait model. The corresponding components for FI, growth, and the efficiency of gain traits were also estimated in a series of 4-trait models (FI, ADG, and MWT were common, with 1 the efficiency of gain trait). The optimization in VCE was done with a quasi-Newton procedure and included setting up the mixed model equations, numerical factorization, solving the triangular system, computing the sparse inverse, and assembling the gradients. The convergence criterion (i.e., the variance of the simplex values) for all runs was 10–9. The following multitrait animal model, in matrix notation, was used to estimate the genetic parameters:
 |
where y = a vector of observations; ß, a, and e = the vectors of fixed, additive genetic, and residual effects, respectively; and X and Z = the known incidence matrices. Fixed effects were sex and selection generation for all traits. The selection generation included the genetic effect of selection and the environmental effect at each generation. Assumptions for random effects were:
 |
where A = the matrix of additive relationships among pigs; Ie = the identity matrix of appropriate order; and 
and 
= the additive genetic and residual variances, respectively. All pigs in the pedigree were used in the analysis to establish the numerator relationship matrix of the pigs.
|
RESULTS |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
. The mean values for FI and ADG were consistent with other reports (Mrode and Kennedy, 1993; Suzuki et al., 2005) in the same breed. Estimated heritabilities for growth traits were moderate to high, ranging from 0.44 to 0.53. Appetite, as measured by FI, was moderately heritable (0.53). The genetic correlations of ADG with FI (0.60) and MWT (0.77) were high and were moderate (0.26) between FI and MWT, whereas the phenotypic correlations among all of them were low to moderate, ranging from 0.14 to 0.51.
|
. Lower values of RFI or FCR indicate better efficiency of an animal. The mean values for RFIphe and RFIgen were close to zero and was negative for RFInut. All 3 estimates of RFI were moderately heritable (0.34, 0.41, and 0.42 for RFInut, RFIphe, and RFIgen, respectively), indicating that variation in RFI of Duroc pigs contains genetic components, which should respond to selection. However, the heritability of FCR was moderate (0.27).
|
. The genetic and phenotypic correlations between RFIphe and RFIgen were greater than 0.95, implying that they may be regarded as the same trait. Corresponding correlations between FCR and different RFI estimates were large but less than 0.90, indicating that FCR should be regarded as a distinct trait. The phenotypic correlations of RFInut with the other 2 RFI estimates were close to unity (>0.95), whereas their genetic correlations were large but less than unity. The positive genetic correlations between RFI and FCR indicate that lower RFI (greater efficiency) was associated with improved FCR.
|
. Feed intake was positively correlated with all measures of the efficiency of gain but was more strongly correlated with RFIphe and RFIgen than with RFInut and FCR. By definition, RFIphe should not be phenotypically correlated with its component traits, and the results confirmed this [i.e., rp with MWT (0.01) and ADG (0.03) were close to zero]. In addition, the results show that RFIgen was genetically independent of MWT (rg = –0.04), whereas a weak genetic correlation was observed between RFIgen and ADG (rg = 0.15). The RFInut was weakly genetically correlated with growth traits. Favorable (negative) genetic and phenotypic correlations were observed between FCR and its component traits. These estimates indicated that as ADG increased, the FCR decreased (greater efficiency).
|
|
DISCUSSION |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
using pen averages of Yorkshire, Landrace, and Duroc boars and to literature averages (0.44) reported by Hutchens and Hintz (1981). However, the present estimate for ADG was greater than the estimate (0.24) by Johnson et al. (1999) in Large White pigs. With Duroc and Landrace pigs, Lo et al. (1992) found a heritability estimate of 0.36 for ADG from 39.5 to 103.6 kg of BW, which was close to the present estimate. Robison and Berruecos (1973) indicated that greater values of heritability for ADG can be expected when pigs are individually fed. Jungst et al. (1981) suggested that differences in estimates of heritability for rate of gain might be due to feeding method (restricted or ad libitum). Differences in source of data and sampling error may be other causes of the differences. Appetite, as measured by FI, appeared highly heritable. Our estimate was comparable to, but slightly greater than, the estimate of 0.45 reported by Mrode and Kennedy (1993) and the estimate of 0.34 for boars fed ad libitum as reported by Cameron et al. (1990). Our study has shown that considerable genetic variation exists in FI, ADG, and MWT in Duroc pigs and that these variations could be exploited in selection programs for the improvement of feed utilization and growth performance.
The genetic and phenotypic correlations between ADG and FI in the current study are in agreement with the findings of Mrode and Kennedy (1993) and Johnson et al. (1999). Mrode and Kennedy (1993) estimated genetic and phenotypic correlations between ADG and FI to be 0.80 and 0.74, respectively. Corresponding correlations estimated by Johnson et al. (1999) for Large White swine were 0.82 and 0.72, respectively. The large estimated genetic and phenotypic correlations between ADG and FI indicate that faster-growing pigs had greater daily feed consumption.
The heritability estimate for FCR in the present study was consistent with previous estimates (Cameron et al., 1990; Mrode and Kennedy, 1993). We are not aware of any published studies on genetic parameters for RFInut in pigs. However, Fan et al. (1995) estimated heritability for RFI (calculated from feeding standard formulas) to be 0.23 in Angus cattle. The heritability estimates for RFIphe and RFIgen obtained in the current study are within the range of previous studies. Foster et al. (1983) found a pooled estimate, across breeds, of 0.30 for RFI adjusted for ADG and backfat, and Haer (1992) reported a heritability of 0.45 for RFI adjusted for ADG and lean growth. Mrode and Kennedy (1993) reported a heritability of 0.30 for RFI when adjusted for ADG and backfat and of 0.38 when adjusted for lean growth rate. The genetic variation found in RFI suggests that opportunities exist to select against RFI as a means of improving the efficiency of gain.
There are few reports in the literature of genetic and phenotypic correlations among different estimates of the efficiency of gain, but those available are in general agreement with the findings of this study. Estimated genetic and phenotypic correlations between RFIphe and RFIgen were greater than 0.95, implying that they can be practically regarded as the same trait. No published literature was found on estimates of the relationships between different measures of the efficiency of gain in pigs. However, Hoque et al. (2006b) reported very large genetic (0.97) and phenotypic (0.98) correlations between RFIphe and RFIgen in beef cattle, concluding that selection for RFIgen would give results similar to selection for RFIphe. The genetic correlations of RFIphe and RFIgen with FCR in the current study were large, which is in agreement with the findings of Herd and Bishop (2000) and Arthur et al. (2001a) in beef cattle.
Calculation of RFI requires estimation of expected FI, which can be obtained through regression or through use of feeding standard formula. Kennedy et al. (1993) showed that when expected FI is obtained by phenotypic regression, RFIphe is expected to be phenotypically independent of the component traits and, when expected FI is obtained by genetic regression, RFIgen is expected to be genetically independent of its component traits. They also mentioned that even when RFI is calculated by genetic regression, there is no guarantee that genetic correlations with component traits will be close to zero. In our study, both MWT and ADG were phenotypically independent of RFIphe, and MWT was genetically independent of RFIgen, whereas there was a weak genetic correlation between RFIgen and ADG. However, when RFI is estimated using expected FI calculated from feeding standard formulas, it is not automatically independent of production traits and is, in fact, usually correlated with these traits, as observed in the current study and in the studies by Fan et al. (1995) and Arthur et al. (2001b) in beef cattle.
The low to moderate, but favorable, negative genetic correlations between FCR and growth traits observed in present study are in agreement with the estimates by Suzuki et al. (2005) and Hoque et al. (2006b). Suzuki et al. (2005) reported the genetic correlation between FCR and ADG to be –0.09 in Duroc pigs, whereas the corresponding correlation was highly negative (–0.52) when estimated by Hoque et al. (2006b) in beef cattle. The estimates from this study suggest that selection for ADG will also lead to slight improvement in FCR, and this has been demonstrated in many selection experiments (Kennedy, 1984). The effect of selection for improved growth rate on appetite is not clear. Some studies have shown a correlated reduction in appetite (Sather and Fredeen, 1978), and others have shown small increases in FI (Ollivier, 1986; Mrode and Kennedy, 1993). Theoretical arguments have been made that selection for ADG under ad libitum feeding would have a nominal effect on FI, at least initially, but that increases in intake might follow in later generations (Fowler et al., 1976). Heritability and genetic correlation estimates from this study suggest that selection for ADG would increase FI. However, direct selection for FCR has not been very effective (Webb and King, 1983), and selection programs that give emphasis to efficiency, rather than growth rate, on ad libitum feeding have led to a reduction in FI (Webb and King, 1983; Smith et al., 1991), which may be detrimental in the long run.
The efficiency of gain is routinely measured as the ratio of FI to product (or the reciprocal). Gunsett (1984) compared the efficiency of direct selection for a 2-component trait with a linear index trait derived from the same 2 components and concluded that the use of a linear index increases selection responses as compared with direct selection on the ratio trait (FCR). Also, selection against FI reduces appetite, which may be undesirable (Ollivier et al., 1990). The RFI, which is a linear index, is derived from the combination of feed consumption and production traits. Results from this study indicate that RFI is heritable. Selection against RFI may provide a means of reducing feed consumption that does not go to a productive purpose. Our study has shown that RFIphe and RFIgen are more favorably and strongly correlated with FI than are FCR and RFInut with FI. For RFI, the weights of the component traits in the selection index are determined only by biological (co)-variances. However, profitability will be maximized when index weights on FI (or RFI), growth, and other traits are determined by both biological and economic parameters, which need to be investigated.
Estimates of heritability reported here indicate that sufficient additive genetic variance exists in RFI, FI, and growth traits, which should lead to response from selection. Genetic correlations indicate that it should be possible to select for reduced RFI without adversely affecting growth performance. The RFI was more heritable than FCR, and FI was more strongly genetically correlated with RFIphe and RFIgen than with FCR and RFInut. Due to these observations, and the favorable nature of RFI (linear index), we concluded that residual feed intake calculated by phenotypic or genetic regression from its component traits should be considered for inclusion in pig breeding programs for genetic improvement of the efficiency of gain.
|
Footnotes |
|---|
2 Corresponding author: azharhoque{at}yahoo.com
Received for publication November 6, 2006. Accepted for publication April 11, 2007.
|
LITERATURE CITED |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
This article has been cited by other articles:
|
|
 |
|
  R. M. Herd and P. F. Arthur Physiological basis for residual feed intake J Anim Sci, April 1, 2009; 87(14_suppl): E64 - E71. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
2P) and additive genetic (