Search
Author
Title
Vol.
Issue
Year
1st Page

Journal of Animal Science - Animal Genetics

Selection to reduce residual feed intake in pigs produces a correlated response in juvenile insulin-like growth factor-I concentration1

 

This article in JAS

  1. Vol. 88 No. 6, p. 1973-1981
     
    Received: Aug 31, 2009
    Accepted: Feb 02, 2010
    Published: December 4, 2014


    2 Corresponding author(s): kbunter2@une.edu.au
 View
 Download
 Share

doi:10.2527/jas.2009-2445
  1. K. L. Bunter 2,
  2. W. Cai,
  3. D. J. Johnston* and
  4. J. C. M. Dekkers
  1. Animal Genetics and Breeding Unit, University of New England, Armidale, New South Wales, 2350, Australia; and
    Department of Animal Science, Iowa State University, Ames 50011

ABSTRACT

Data from a selection experiment for residual feed intake (RFI) were used to estimate genetic correlations between measures of efficiency and performance traits with juvenile IGF-I, and to demonstrate direct and correlated responses to selection. The heritability of IGF-I was 0.28 ± 0.06 and genetic correlations of IGF-I with feed intake (0.26 ± 0.17), backfat (0.52 ± 0.11), and feed conversion ratio (0.78 ± 0.14) were moderate to large. The estimated and realized genetic correlations between RFI and IGF-I were 0.63 ± 0.15 and 0.84. In contrast, genetic correlations between IGF-I and lifetime or test period growth did not differ (P > 0.05) significantly from zero (0.06 ± 0.14 and −0.19 ± 0.14). Selection for decreased RFI produced a direct response in RFI, as expected, and was accompanied by downward correlated responses in ADFI, juvenile IGF-I, backfat, and growth traits, listed in order of decreasing relative magnitude, and an increased loin muscle area. The correlated response in IGF-I to selection on RFI demonstrates that this physiological measure is genetically associated with efficiency, and is thus useful as an early information source to estimate genetic merit for efficiency before performance testing. Decreased juvenile IGF-I is associated with leaner, more efficient animals.



INTRODUCTION

Residual feed intake (RFI) is a measure of production efficiency in growing animals. Residual feed intake is defined as the amount by which actual feed intake differs from expected intake, given the predicted requirements for maintenance and production output of an animal. The definition of production output typically includes growth (cattle) and composition traits (pigs), but it can also include other forms of output, such as egg or milk production, in other species. Different trait definitions of RFI can obviously result in different outcomes under selection.

Within the range of possible definitions for RFI, and with all other factors held constant, differences between individuals in their RFI are thought to be related to variability in factors other than growth and body composition (or production output), such as feeding behavior, nutrient digestion, maintenance requirements or activity, and energy homeostasis or partitioning. The current lack of understanding about the biological consequences of selection on RFI was part of the motivation for development of the selection line at Iowa State University (ISU; Cai et al., 2008) used in this study. The ISU selection line quantifies correlated changes in production traits under single-trait selection for RFI.

Estimates of genetic parameters from previous studies (Luxford et al., 1998a,b; Lahti et al., 2001; Bunter et al., 2002, 2005) have suggested that circulating concentrations of juvenile IGF-I can provide information on the genetic merit of individuals for efficiency and some related production traits. Insulin-like growth factor-I is a naturally occurring polypeptide present in the circulating blood supply. It is known to influence growth and development during the postnatal period (Hossner et al., 1997). This physiological indicator, which can be measured early in life, is potentially an informative selection criterion in breeding programs that target improved efficiency.

A single-generation divergent selection experiment for juvenile IGF-I generally supported the differences implied by prior estimates of genetic parameters from the same population (Luxford et al., 1998b). However, Cameron et al. (2003) was able to replicate some of these associations only within specific selection lines. Cameron et al. (2003) argued that although estimates of genetic parameters from large populations may be precise, validation of predicted responses to selection is not achieved. They suggested that a better strategy to evaluate candidate traits for physiological prediction of genetic merit would be by illustrating a correlated response from selection on a production-oriented trait. The purpose of this study was therefore to examine the correlated response in juvenile IGF-I and production traits resulting from selection solely on RFI.


MATERIALS AND METHODS

Experimental procedures for this study were approved by the ISU Institutional Animal Care and Use Committee.

Performance Recording and Breeding Strategies Used in the Selection Experiment

The design and conduct of the selection experiment for RFI was described in detail by Cai et al. (2008). Commencing in 2001, the experiment was structured as a line of Yorkshire pigs selected for reduced RFI (Select line) over 5 generations compared with a randomly selected line (Control line). Performance testing in the Select line consisted of recording individual feed intake in group pens fitted with FIRE (Osborne Industries Inc., Osborne, KS) electronic feeders, after a 1-wk acclimation period. Serial recording of BW, ultrasonic backfat, and loin muscle area was performed until pigs reached a target fixed end BW of approximately 115 kg. Boars and gilts generally commenced testing at 30 to 40 kg. For pigs recorded for feed intake, removal from test occurred when the target BW was reached on an individual basis until only 3 pigs per pen remained, which were then removed together. All other pigs were removed from test on a pen basis when the average pen BW reached the target BW, resulting in significant variability between individuals in both off-test BW and age. The significance of the performance test system for regressions involving BW was considered when developing models for analyses.

In each generation, up to 96 males (generally 2 per litter) arising from parity 1 litters in the Select line were recorded for feed intake and were selected on the basis of their EBV for RFI only. Gilts were selected on their EBV derived from univariate genetic evaluation of RFI data from relatives. Selection of full-sib brothers was avoided, and there was also a limit on the number of gilts selected per litter to reduce rates of inbreeding. Targets were to produce approximately 50 litters from 12 to 15 selected boars per generation in the Select line and approximately 25 litters from 9 to 10 boars per generation in the Control line. After selection of animals from parity 1 litters, additional performance data from full- and half-sibs was produced from repeat mating to generate parity 2 litters. These additional data were used to increase the accuracy of evaluation for response in RFI of the Select line. In contrast, animals for the Control line were selected at random and only limited feed intake recording was conducted to enable the line contrast for feed intake. Whereas Control line animals were recorded for feed intake in generations 3 and 4 (females) or 5 (males) only, more extensive performance recording was conducted in this line for growth and backfat traits.

Genetic evaluation procedures and the motivation behind models used for estimating breeding values for RFI were described by Cai et al. (2008). Essentially, EBV for RFI were generated using all available historical data under a single-trait animal model for ADFI, with the additional random effects of litter and pen cohort. Fixed effects included on-test group, sex, and the linear covariates of on-test age, on-test BW, and off-test BW, along with ADG and backfat at the end of test, adjusted to an on-test age of 90 d and an end-test BW of 115 kg, respectively. Linear covariates were fitted separately for each parity and generation, effectively allowing the contributions of production traits toward RFI of selection candidates to differ by generation. Metabolic mid-weight was not included as a linear covariate for defining RFI, in contrast to the common definition of RFI in other studies. However, Cai et al. (2008) reported a correlation of 0.98 between EBV including or excluding metabolic mid-weight from the above model factors, indicating that maintenance requirements associated with metabolic mid-weight were already accounted for in their models by the combination of BW and BW gain covariates fitted.

Performance records used in this study included lifetime (LADG; kg/d) and on-test ADG (TADG; kg/d), ADFI (kg/d), backfat depth at the P2 site, which is approximately 6.5 cm from the dorsal midline, at the level of the posterior edge of the head of the last rib (BF; mm), and loin muscle area (LMA; cm2), measured with real-time ultrasound (Aloka Inc., Wallingford, CT) at the end of test. Variation in birth weight was not accounted for in the calculation of LADG. Test ADG was defined as the linear regression of BW on days. The number of BW recorded while on test varied between animals and generations. The ADFI for individual animals was derived using procedures described by Casey et al. (2005) and Cai et al. (2008). Feed conversion ratio (FCR; kg/kg) was ADFI/TADG, whereas RFI (kg/d) for this study was modeled slightly differently from that used by Cai et al. (2008), as described in the section on parameter estimation.

Testing for juvenile IGF-I commenced in generation 2. These data were not used for genetic evaluation or selection decisions so that the independent correlated response in IGF-I to selection for RFI could be investigated. Blood for IGF-I testing was obtained via venipuncture of the orbital plexus on the medial side of the eye using a glass capillary tube. This was then transferred to a blood spot card, air-dried, and stored at −80°C before laboratory analyses. Samples were analyzed by Primegro Ltd. (Corowa, Australia) using a commercially available ELISA assay (Diagnostic Systems Laboratories Inc., Webster, TX). This kit is a nonextraction ELISA that uses an enzymatically amplified 2-step sandwich-type immunoassay. The kit has been validated to quantify IGF-I concentration in human plasma with an intraassay CV of <10% and sensitivity for minimum detection of 0.01 ng/mL. Blood sampling in all piglets occurred after weaning, with samples in males generally collected before castration for parity 1 litters and after castration for parity 2 litters. Juvenile IGF-I data were included in the analyses only if weaning date was known and piglets were blood sampled at ≤42 d of age. The targeted age for the IGF-test was 35 d of age (range 33 to 42 d of age).

Any animal weaned within the first week after birth that survived to obtain subsequent IGF-I or performance records was excluded from analyses. Age at weaning after this edit was quite variable (7 to 40 d) because of the farrowing schedule at the farm, with a mean of 22 d. The remaining data for all traits were examined for their distributional properties using the UNIVARIATE procedure (SAS Inst. Inc., Cary, NC). Suitable thresholds for identifying outliers were established by comparing schematic box plots across generations, lines, and sex. Outliers were generally defined as exceeding 1.5× the interquartile interval across relevant data subsets. Feed intake and associated test period data were also deleted for animals with a very short test period (≤21 d), if they were older than 180 d of age or more than 100 kg at the beginning of testing, and if off-test BW were outside the range of 65 to 155 kg. Records for these animals were generally unrepresentative of the normal performance testing procedures or cohort data. After these limits and edits were imposed, individual records for FCR exceeding 4 kg/kg, or LMA <15 cm2, were also deleted as outliers. For most traits, editing for outliers reduced the quantity of data used for analyses by approximately 2.5%; reductions ranged from 1.1% for IGF-I to 4.3% of original data for test daily BW gain.

Parameter Estimation

The data and models for estimating genetic parameters and evaluation of (correlated) response to selection in the current study are specified in Table 1 and were slightly different from the models used for genetic evaluation and selection during the selection experiment (Cai et al. 2008). Specifically, in this study, metabolic mid-weight was explicitly included as a linear covariate for defining RFI, and the necessity of fitting all other linear covariates, either across or within generations, was then reassessed, retaining only those effects that were significant at 5% in the complete model (Table 1). Metabolic mid-weight was defined as the average of beginning and end test BW, expressed as kilograms0.75. For UNIVARIATE analysis of RFI, records for RFI were not generated explicitly but were modeled by including covariates for ADG, BF, and metabolic weight in the analysis of feed intake (model [f]; Table 1), which is equivalent to estimating parameters for predicted RFI based on regression coefficients from the same model. Simple correlations between RFI phenotypes from models used in this study and those of Cai et al. (2008) ranged from 0.91 to 0.99 within generation and, when combined with almost identical trait variation, demonstrate that these alternatives resulted in very similar trait definitions for RFI.

Table 1.

Please see the pdf to view this table.

 

Starting test date is comparable with the “Group” term of Cai et al. (2008), indicating groups of animals that commenced testing together, and this effect also accounted for seasonal and farrowing batch effects on traits such as LADG. Animals from the Select and Control lines were managed as contemporaries and analyzed jointly. Line was not included as a fixed effect for any trait because both the Select and Control lines were derived from a common base population. Data from gilts, boars, and barrows were analyzed jointly and were assumed to have the same genetic parameters.

Estimates of genetic parameters and EBV were obtained using REML procedures under an animal model using ASREML (Gilmour et al., 1999). Univariate analyses were used to derive fixed effects models and to estimate heritabilities. Genetic correlations between traits were estimated from bivariate analyses using the same trait models, except for RFI. Because of the colinearity problems described by Cai et al. (2008) for bivariate analyses of RFI with other traits, an RFI phenotype was generated for each individual that had complete data on ADFI and covariates, using the estimates of regression coefficients from the univariate analysis of RFI for model [f] in Table 1. Terms for covariates were then removed from the model for the resulting RFI phenotype in the bivariate analyses.

Random effects included animal and litter effects for all traits except ADFI and RFI, for which litter effects were not significant (P > 0.05) based on a likelihood ratio test. Pen cohort was fitted as an additional random effect for all traits. The mixed model is represented as y = Xb + Z1a + Z2c + Z3pc + e, where y is the vector of observations; X, Z1, Z2, and Z3 are incidence matrices relating records to effects b, a, c, and pc; b is the vector of solutions for fixed effects; a, c, and pc are vectors of solutions for additive genetic common litter and pen cohort effects and e is the vector of residuals I1, I2, and I3 are identity matrices, and A is the matrix describing additive genetic relationships between animals. The pedigree contained 361 sires and 782 dams; generations were discrete. Twenty-three sires and 44 dams were common parents across the Select and Control lines in generation −1, for animals recorded in generation 0.

Line Differences

For all traits except RFI, line differences were calculated as the difference in average EBV between lines by generation, using EBV from bivariate analyses of the trait and the predicted phenotype for RFI, to account for selection on RFI. Because selection was directly on EBV for RFI, line by generation differences for RFI were estimated using EBV from the univariate analysis of feed intake using model [f] in Table 1. Significance of line differences was approximated by pairwise comparisons of mean EBV by line and generation, with Bonferroni correction for multiple comparisons. Although the statistical assumptions of the least squares approach to general linear modeling are not strictly met in the comparison of mean EBV, the significance tests are a useful approximation. The difference between lines was divided by the relevant trait genetic SD (GSD) to illustrate the relative response for each trait.


RESULTS AND DISCUSSION

Characteristics of the Data and Heritability Estimates

Characteristics of the edited performance data are shown in Table 2, along with univariate estimates of heritabilities, common litter and pen cohort effects, and the phenotypic SD. Compared with growth and composition traits, which had CV in the range of 11 to 29%, the variability of juvenile IGF-I was generally large (CV = 46%). However, systematic effects were significant contributors to variation in IGF-I, explaining 44% of the variation in this trait. Average ages at the beginning and end of performance testing were 94 and 194 d, with CV of 20 and 8%, respectively. Corresponding average BW were 40.2 and 112.6 kg, with CV of 36 and 10%.

Table 2.

Please see the pdf to view this table.

 

In this study, the systematic effects of test group and sex accounted for approximately 22% (LADG, TADG, and ADFI) of the variation in the raw data. Based on incremental additions of each covariate to the systematic effects model, variation in metabolic mid-weight accounted for an additional 28% of the variation in feed intake, whereas BW and BF gain while on test, in combination, accounted for an additional 18 to 19% of variation in intake. The percentage of variation in feed intake explained by systematic effects from the full model (model [f]; Table 1) was 60%. Cai et al. (2008), who fitted covariates by generation and parity, found that 66% of the phenotypic variation in feed intake was accounted for by covariates. However, some of this variation was likely repartitioned from test group and sex effects. Using models containing BW gain and BF thickness as covariates, Gilbert et al. (2007) found that these covariates accounted for 66 to 76% of the raw data variation in feed intake.

Estimates of heritability and common litter effects for IGF-I (Table 2) were similar to those observed previously (Bunter et al., 2005). Parameter estimates for the remaining traits were generally consistent with those reported by Cai et al. (2006, 2008) for the same population but estimated using a subset of these data; any differences are consistent with the use of different data (complete through to generation 5 in this study), data editing procedures, and the analytical models used. The estimates of phenotypic variation for all traits were larger in this study compared with those estimated by Cai et al. (2008), who used only Select line data to generation 4. In this study, the heritability estimate for RFI was greater when litter was excluded, supporting a negative sampling correlation between additive and litter effects attributable to confounding, which is commonly observed in pig data, or an upward bias in the heritability estimate when litter is not included in the model. However, the litter effect was not significant for RFI and was excluded from the model for RFI on this basis. Overall, parameter estimates for performance traits were generally similar to those reported elsewhere (Clutter and Brascamp, 1998), and the heritability for RFI was consistent with reports from other studies examining RFI in pigs (Mrode and Kennedy, 1993; Nguyen et al., 2005; Gilbert et al., 2007; Hoque et al., 2009), although toward the larger end of the published range of estimates.

Genetic Correlations Between IGF-I, RFI, and Performance Traits

Estimates of correlations between IGF-I and ADFI or RFI were obtained from models excluding common litter effects for the latter 2 traits because they were not significantly different (P > 0.05) from zero (Tables 3 and 4). For all traits, estimates of trait heritabilities, common litter or pen cohort effects, where applicable, and variances from bivariate analyses (not presented) were very similar to those obtained from univariate analyses.

Table 3.

Please see the pdf to view this table.

 
Table 4.

Please see the pdf to view this table.

 

For juvenile IGF-I, estimates of genetic correlations were moderate to strong and positive with BF, RFI, and FCR (0.52, 0.63, and 0.78) but were smaller with ADFI (0.26; Table 3). Estimates of genetic correlations for juvenile IGF-I with LADG or TADG were relatively weak and were not significantly different (P > 0.05) from zero (0.06 and −0.19). The negative genetic correlation of IGF-I with test growth rate, although not significantly different from zero, combined with strong positive correlations of IGF-I with BF and FCR, are consistent with decreased IGF-I and improved efficiency being associated with later maturing animals with reduced fat deposition. The strong genetic correlation between IGF-I and RFI suggests that associations exist between IGF-I and efficiency beyond those solely resulting from changes in BF and ADG, although this is difficult to substantiate definitively because the SE of estimates are large.

Phenotypic correlations between IGF-I and performance traits were consistent in direction with genetic correlations but were generally much smaller in magnitude because of the general absence of significant residual correlations between these trait groups, which were measured at very different ages (Table 3). Estimates of genetic and phenotypic correlations were consistent with the average estimates reported by Bunter et al. (2005) from other populations. However, the genetic correlation between ADFI and IGF-I was weaker in this study, whereas the correlation between IGF-I and TADG was stronger. There is currently no other literature reporting parameter estimates between juvenile IGF-I and feed efficiency in pigs. Genetic correlations from this study indicate that selection for decreased RFI, or efficiency as defined by FCR, should produce a correlated downward response in IGF-I, and vice versa.

Estimates of genetic correlations of RFI with performance traits are shown in Table 4. Genetic correlations of IGF-I or RFI with performance traits (Tables 3 and 4) were generally comparable in direction, as would be expected given the relatively strong genetic correlation between IGF-I and RFI (Table 3). The exception was for TADG, which had a negative genetic correlation with IGF-I (−0.19 ± 0.14), suggesting that more efficient animals had greater TADG (Table 3) but a positive genetic correlation with RFI (0.24 ± 0.12), suggesting, less plausibly, that reduced growth is associated with better efficiency (Table 4). Although neither of these correlations differed significantly (P > 0.05) from zero, they did differ significantly (P < 0.05) from each other. However, given the magnitude of the genetic correlation between IGF-I and RFI (0.63 ± 0.15), this difference is within possible bounds. Further, as was noted previously (Kennedy et al., 1993; Mrode and Kennedy, 1993), because RFI is defined using phenotypic and not genetic regressions, estimates of genetic correlations for RFI with traits that contribute to its prediction may be nonzero. Estimates of genetic correlations between RFI and TADG or BF were very similar to comparable traits reported by Mrode and Kennedy (1993) and Cai et al. (2008). For the latter study, the exception was the correlation between RFI and BF, which was positive (0.20 ± 0.11) in the current study but negative (−0.14 ± 0.16) in the study by Cai et al. (2008). These estimates do, however, not differ significantly from zero or from each other. Significant residual, pen cohort, and phenotypic correlations were evident for RFI with ADFI and FCR, although phenotypic correlations between RFI and BF or TADG did not differ significantly from zero, as expected (Kennedy et al., 1993).

Correlated Responses to Selection for Decreased RFI

Correlated responses in performance traits to selection for decreased RFI can be quantified by line differences in EBV at generation 5 (Table 5). Pedigree and data were available before generation 0, from which animals were selected as parents for the 2 lines. Although the expectation was for no trend in any trait under random selection, the mean EBV differed significantly (P < 0.05) between G5 and G0 in the unselected Control line for FCR. The potential causes of the positive trend in EBV for FCR in the Control line are not obvious, but might simply reflect problems inherent with the analysis of ratio traits. In addition, compared with the Select line, relatively few animals (approximately 200) within the Control line have data for FCR, in particular because feed intake was recorded only from generation 3 onward in this line. Therefore, it is possible that the genetic trend for FCR within this line was less accurate. Further, analysis of FCR can be problematic unless the test length is constant because the relative change in accuracy of recording the components of FCR (ADFI and TADG) differs with changing test length (Giles et al., 2007). Performance recording to a target end BW, as occurred in these data, will give rise to a negative association between BW gain on test, test length, and the number of data points used to estimate TADG, which has consequences for the accuracy of the data and resulting parameter estimates.

Table 5.

Please see the pdf to view this table.

 

Single-trait selection for decreased RFI was effective in generating a significant difference between lines in RFI. Significant correlated responses were generally also observed in the Select line for traits that contribute to differences in efficiency (i.e., feed intake, growth rates, and body composition, as described by BF). The correlated response in ADFI was expected, whereas accounting for growth and body composition to determine RFI was intended to limit correlated responses in these traits. However, although phenotypic correlations between RFI and TADG or BF did not differ (P > 0.05) from zero (Table 4) as expected, because of the model used for RFI, genetic correlations may be nonzero because the adjustments for growth and BF to compute RFI were at the phenotypic level. The direction of correlated responses was consistent with estimates of genetic correlations (Table 4). A significant correlated response occurred in juvenile IGF-I under selection for RFI.

The relative response for each trait is illustrated in Figure 1, expressed as the difference of the average EBV for the Select line from the Control line, rescaled to GSD. The greatest response was obtained for RFI, which, as explained previously, is a simple reparameterization of the Select line selection criterion described by Cai et al. (2008). Trends for RFI appear curvilinear and the pattern of correlated responses in other traits also tended to differ over time. This possibly reflects the genetic evaluation strategy of Cai et al. (2008), which allowed differential contributions of production traits toward predicted RFI in each generation. This potentially alters the trait definition each generation, which has implications for the accuracy of EBV. The consistent decline in response may also simply reflect reductions in genetic variance attributable to the Bulmer effect or inbreeding, and potentially a decline in the accuracy of recording (e.g., a declining number of serial measurements) over time.

Figure 1.
Figure 1.

Difference between mean EBV for Select and Control lines for each trait (expressed in genetic SD for each trait). LADG = lifetime ADG; TADG = test ADG; BF = backfat; LMA = loin muscle area; RFI = residual feed intake; FCR = feed conversion ratio (see text for descriptions of traits).

 

The greatest correlated response was observed for feed intake (Figure 1). This outcome is to be expected because RFI was the selection criterion for the Select line and RFI is more highly correlated with feed intake than with growth traits (Table 4; Cai et al., 2008). The change in ADFI (–1.11 GSD units by generation 5) was slightly less than for RFI (−1.13 GSD), whereas downward trends for BF, LADG, and TADG in the Select line were of smaller magnitude (−0.65 for TADG to –0.80 for BF, in GSD units). The trend for FCR mirrored that for RFI but was much smaller in magnitude (−0.49 GSD units). This demonstrates that FCR and RFI are not genetically identical traits for describing efficiency; the genetic correlation between these traits was 0.70 ± 0.09. The genetic correlation between ADFI and FCR was also less (0.20 ± 0.14) than that between RFI and ADFI (0.52 ± 0.12; Table 4). However, from additional analyses in which the model for FCR was altered to accommodate variation in test length, the genetic correlation between RFI and FCR increased to 0.87 ± 0.04 and that between ADFI and FCR increased to 0.58 ± 0.11, demonstrating the importance of test length in particular on outcomes for FCR, largely through its greater impact on the accuracy of assessing growth rate relative to feed intake (Giles et al., 2007). Under this model for FCR, the correlated response in FCR to selection on RFI also increased in magnitude to −0.63 GSD.

The correlated response for IGF-I was –1.07 GSD units by generation 5 (Figure 1), which was in the expected direction given the estimate of the genetic correlation with RFI, and was of large magnitude. The correlated response in IGF-I to selection on RFI equates to a realized genetic correlation of 0.84, calculated using parameters from the bivariate analysis asThe realized estimate is larger than the direct estimate of the genetic correlation between these traits of 0.69 ± 0.15 (Table 4) but was within a SE of the REML estimate.

Response to selection on RFI and correlated responses in production traits were greater than were observed from the divergently selected lines for RFI in the French lines of Gilbert et al. (2006) at generation 3. In the French study, between-line differences in test ADG and BF were not significant, although they demonstrated a consistent divergence between lines over generations for BF. The most recent results for the French population support, on average, moderate positive genetic correlations between the derived RFI and BF traits in particular (Gilbert et al., 2007), although line differences were not reported. Thus, it is possible that the lack of significant correlated responses in production traits reported by Gilbert et al. (2006) was simply due to the reduced accuracy or intensity of selection, combined with fewer generations of selection. The ISU Select line data also showed a substantially larger relative and absolute change in feed intake under selection compared with the data from France, where in relative magnitude, the responses for RFI, DFI, and FCR, when expressed in SD units, were more similar to each other (Gilbert et al., 2006). These studies were conducted in different populations, but the differences observed likely also arise from the different definitions for the selection criterion. These include differences in the age and BW range over which RFI components were recorded, and therefore the relative contributions of maintenance, BW gain, and changes in body composition to RFI, along with the use of constant (French experiment) or altering definitions of RFI over time (ISU experiment) to form values for the selection criterion. Such differences will be reflected in estimates of genetic parameters.

Conclusions

Estimates of genetic correlations and selection line differences to quantify correlated responses provide supporting evidence for the previously reported genetic correlations between juvenile IGF-I and performance traits, estimated in different populations (Bunter et al., 2005). In a line selected solely for improved efficiency through reduced RFI, a correlated response in the expected downward direction was observed for juvenile IGF-I, confirming that it is a good physiological indicator of genetic merit for economically important efficiency traits, particularly because it is measured early in the life of an animal.

 

References

Footnotes


Comments
Be the first to comment.



Please log in to post a comment.
*Society members, certified professionals, and authors are permitted to comment.