Genetic correlations between two strains of Durocs and crossbreds from differing production environments for slaughter traits

Genetic correlations between two strains of Durocs and crossbreds from differing production environments for slaughter traits B. Zumbach, I. Misztal, S. Tsuruta, J. Holl, W. Herring and T. Long J Anim Sci 2007.85:901-908. doi: 10.2527/jas.2006-499 originally published online Dec 18, 2006; The online version of this article, along with updated information and services, is located on the World Wide Web at: http://jas.fass.org/cgi/content/full/85/4/901 www.asas.org

Genetic correlations between two strains of Durocs and crossbreds from differing production environments for slaughter traits B. Zumbach, 1 * I. Misztal,* S. Tsuruta,* J. Holl, W. Herring, and T. Long *Department of Animal and Dairy Science, University of Georgia, Athens 30602; and Smithfield Premium Genetics Group, Rose Hill, NC 28458 ABSTRACT: The aim of this study was to estimate the genetic correlations between 2 purebred Duroc pig populations (P1 and P2) and their terminal crossbreds [C1 = P1 (Landrace Large White) and C2 = P2 (Landrace Large White)] raised in different production environments. The traits analyzed were backfat (BF), muscle depth (MD), BW at slaughter (WGT), and weight per day of age (WDA). Data sets from P1, P2, C1, and C2 included 26,674, 8,266, 16,806, and 12,350 animals, respectively. Two-trait models (nucleus and commercial crossbreds) for each group included fixed (contemporary group, sex, weight, and age), random additive (animal for P1 and P2 and sire for C1 and C2), random litter, and random dam (C1 and C2 only) effects. Heritability estimates (±SE) for BF were 0.46 ± 0.04, 0.38 ± 0.02, 0.32 ± 0.02, and 0.33 ± 0.02 for P1, P2, C1, and C2, respectively. Heritability estimates for MD were 0.31 ± 0.01, 0.23 ± 0.02, 0.19 ± 0.01, and 0.12 ± 0.01 for P1, P2, C1, and C2, respectively. The estimates for WGT and WDA were 0.31 ± 0.01, 0.21 ± 0.02, 0.16 ± 0.01, and 0.18 ± 0.01 and 0.32 ± 0.01, 0.22 ± 0.02, 0.16 ± 0.01, and 0.19 ± 0.01, respectively. Genetic correlations between purebreds and crossbreds for BF were 0.83 ± 0.09 (P1 C1) and 0.89 ± 0.05 (P2 C2), for MD 0.78 ± 0.05 (P1 C1) and 0.80 ± 0.08 (P2 C2). For WGT and WDA, the correlations were 0.53 ± 0.08 (P1 C1), 0.80 ± 0.10 (P2 C2), and 0.60 ± 0.07 (P1 C1) and 0.79 ± 0.09 (P2 C2), respectively. (Co)variances in crossbreds were adjusted to a live BW scale. Compared with purebreds, the genetic variances in crossbreds were lower, and the residual variances were greater. Sire variances in crossbreds were approximately 20 to 30% of the animal variances in purebreds for BF and MD but were 13 to 25% for WGT and WDA. The efficiency of purebred selection on crossbreds, assessed by EBV prediction weights, ranged from 0.43 to 0.91 for line 1 and 0.70 to 0.92 for line 2. When nucleus and commercial environments differ substantially, the efficiency of selection varies by line and traits, and selection strategies that include crossbred data from typical production environments may therefore be desirable. Key words: crossbreds, genetic correlation, genotype environment interaction, pig, plasticity, variance component 2007 American Society of Animal Science. All rights reserved. J. Anim. Sci. 2007. 85:901 908 doi:10.2527/jas.2006-499 INTRODUCTION Structured crossbreeding efficiently exploits additive and nonadditive effects in commercial swine production. Although economic importance is focused on crossbred performance, the selection is usually based on purebred performance. The efficiency of this selection depends on the genetic correlation between purebreds and crossbreds (r pc ). Crossbreds are usually raised in environments of lower quality than those of purebreds concerning hygiene status and space per pig. These factors affect feed 1 Corresponding author: birgit@uga.edu Received July 25, 2006. Accepted December 7, 2006. intake and growth (Schinckel et al., 1999; Wolter et al., 2002). Subsequently, low r pc may be due to not only purely genetic factors but also the genotype environment interaction. Genotypes differ in their plasticity, e.g., levels of adaptation to specific environments (Via et al., 1995; Merks et al., 2005). A highly plastic line would perform well in an optimal environment but possibly poorly in environments that are less than optimal. Less plastic lines would perform similarly across many environments, although their performance may be less than that of high-plasticity lines in an optimal environment. Highly plastic lines would be desirable in locations in which production settings for purebreds and crossbreds are similar, e.g., in Northern Europe. A less plastic line could be more productive in locations with large 901

902 Zumbach et al. Table 1. Number of records, unadjusted means, and SD in the genetic groups No. of Genetic group 1 observations Mean SD Age at measurement, d P1 26,674 172.2 5.0 P2 8,266 171.2 5.0 C1 16,806 197.7 9.5 C2 12,350 194.9 9.2 Weight at measurement, 2 kg P1 26,674 114.2 11.8 P2 8,266 120.3 12.9 C1 16,806 89.6 10.8 C2 12,350 89.1 11.2 Weight per day of age, 2 g/d P1 26,674 664 67 P2 8,266 703 75 C1 16,806 454 55 C2 12,350 458 56 Backfat, 3 mm P1 25,630 13.3 3.1 P2 8,258 17.2 4.3 C1 16,806 19.4 4.2 C2 12,348 20.6 4.5 Muscle depth, 3 mm P1 25,631 46.2 5.5 P2 8,259 43.9 5.5 C1 16,806 58.3 7.3 C2 12,350 56.3 7.3 1 P1 = purebred Duroc line 1; P2 = purebred Duroc line 2; C1 = P1- sired crossbred; and C2 = P2-sired crossbred. 2 Live BW for purebreds; carcass weight for crossbreds. 3 Ultrasonic measurements on live purebreds; optical measurements on the carcasses of the crossbreds while warm. environmental changes, e.g., large farms in the United States. An approximate indication of plasticity is provided by r pc. The value of this correlation for traits of interest can also be used in making decisions regarding appropriate selection strategies, i.e., pure-line selection, reciprocal recurrent selection, or a combination of both (Wei and van der Steen, 1991; Wei et al., 1991; Wei and van der Werf, 1994). Selection for plasticity is possible (Scheiner, 2002). The goal of this study was to estimate r pc for growth, fatness, and muscling between 2 Duroc sire lines and their corresponding terminal crossbreds. Data MATERIALS AND METHODS Data from 2 Duroc nucleus lines (P1, P2), including pureline offspring and their commercially raised terminal crossbred progeny from Landrace (LR) Large White (LW) dams [C1: P1 (LR LW); C2: P2 (LR LW)], were available (Smithfield Premium Genetics Group). Nucleus data were collected from July 2003 through April 2006 and crossbred data from April 2004 through April 2006. Animal Care and Use Committee Table 2. Number of sires and dams in the genetic groups Genetic group 1 Sires, no. Dams, no. P1 263 2,319 P2 145 898 C1 170 2,111 C2 83 1,786 1 P1 = purebred Duroc line 1; P2 = purebred Duroc line 2; C1 = P1- sired crossbred; and C2 = P2-sired crossbred. approval was not obtained for this study because the data were obtained from an existing database. Nucleus animals were produced on 4 farrow-to-finish farms. On 2 of them (50% of the data), P1 and P2 were equally represented. On farm 3, only P1 animals were produced; on farm 4, 95% of P1 and 5% of P2 animals were kept. There was no occurrence of porcine reproductive and respiratory syndrome virus, Actinobacillus pleuropneumonia, ormycoplasma hyopneumoniae on these farms for the duration of this project. During this study, these nucleus farms provided 25% more finishing floor space to the pigs than their commercial counterparts. Commercial crossbreds were raised on two, 1,200- sow farrow-to-finish farms. The crossbreds C1 and C2 were represented on both farms at 58 and 42%, respectively. These farms were known to be affected by porcine reproductive and respiratory syndrome virus, A. pleuropneumonia, and M. hyopneumoniae, as confirmed by serology and diagnostic testing. Pig space was typical of commercial production settings. Typical and similar corn-soybean diets were used at all sites. Feeding was ad libitum for all animals. Measurements on pureline animals were taken at an average age of 172 d. These included live BW (WGT), ultrasound backfat (BF) at the 10th rib, and ultrasound muscle depth (MD; Aloka model 500V real-time ultrasound, Corometrics Medical Systems, Wallingford, CT). Weight per day of age (WDA) was calculated as the ratio between weight and age at measurement. Crossbred animals were slaughtered at an average age of 196 d. Carcass measurements by Fat-O-Meater (SFK Technology, A/S, Herlev, Denmark) included carcass weight (WGT), BF, and MD. Carcass WDA was calculated as carcass weight divided by age at slaughter. Only the pedigreed data from P1, P2, C1, and C2 were considered in the analyses (Table 1). There were no pedigree data on the LR LW crossbred dams. The pedigree file contained 27,171 and 9,802 animals for P1 and P2, respectively, of which 1,519 and 1,513 parents were without records. The number of sires and dams for each genetic group is presented in Table 2. Statistical Analyses The 2-trait model was

Correlations between Durocs and crossbreds 903 y P = X P β P + Z P a P + W P l P + e P, y C = X C β C + Z C s C + U C d C + W C l C + e C, d C = d ac + d ec, and a C = s C + d ac, where P indicates the Duroc line; C indicates the corresponding crossbred line; y = a vector of observations; and β = a vector of fixed effects including the contemporary group (farm-finisher barn-year-week) and sex; weight and age at measurement were included as covariables. For WGT as an analyzed trait, weight was omitted as a covariable, and for WDA, weight and age were omitted as covariables. Additionally, a P = a vector of additive genetic effects of the animal for P; s C = a vector of additive genetic effects of the sire for C; and d = a vector of dam effects composed of dam additive genetic effects (d ac ) and dam environmental effects (d ec ). Further, l and e = vectors of birth litter and residual effects, respectively, and X, Z, U, and W = the appropriate incidence matrices. The expectations of a, s, d, l, and e were assumed to be E a P s C d C l P l C e P e C 0 0 0 = 0. 0 0 0 Assuming that the differences between breeds were negligible and that the environments and traits were the same in both P and C, the following applies: Var(s C )=¹ ₄ Var(a C ), Var(d ac )=¹ ₄ Var(a C ), and Var(Φ) =¹ ₂ Var (a C ), where Φ = the Mendelian sampling. Because Φ is not part of the model for the crossbreds, it becomes part of the residual. The variances were assumed to be a Var P s C = σap 2 σ apsc σsc 2 σ apsc A, where A = the numerator relationship matrix and I = appropriate identity matrices of appropriate dimensions m, n, q, r, and t. No maternal effects were included in the model based on preliminary analyses. Heritabilities in crossbreds were calculated assuming the additive variance was 4 the sire variance. Analyses were performed using AIREMLF90 (Misztal et al., 2002). Based on only purebred information, the breeding values of the sires used on the commercial farms can be predicted as follows: u c u p = u p σ2 c r σp 2 pc, with σ2 c r σp 2 pc representing the weight, where u c = the EBV of the crossbred animal; u p = the EBV of the purebred animal; σ 2 c = the additive genetic variance of the crossbred population; σ 2 p = the additive genetic variance of the purebred population; and r pc = the genetic correlation between purebreds and crossbreds. The growth traits in purebreds and crossbreds were different, i.e., WGT in the purebreds and carcass weights in crossbreds. Following Johnson et al. (2004) and Schinckel et al. (2001), we assumed that the carcasses were 75% of WGT and that the SD of the carcasses were 83% of the SD of live BW. Subsequently, variances at the WGT level were assumed to be 1.45 of those measured at the carcass level, and the prediction formulas above need to be multiplied by 1.2. RESULTS Unadjusted means are presented in Table 1. On average, commercial animals were 24 to 26 d older at measurement than the nucleus animals. Assuming 75% dressing percentage (Boggs et al., 2006), the average carcass weight of C1 and C2 adjusted to WGT would be 119 kg, which is on the same level as P2 (120.3 kg). Thus, adjusted WDA was lower in the crossbreds than in the purebreds. However, compared with the purebreds, the crossbreds exhibited more BF and increased loin depth. Backfat l Var P l C = I m σlp 2 0 Variance components for BF are presented in Table 0 I n σlc 2, 3. The additive genetic variance of C1 (0.89), which is a sire variance, was about 25% of the additive variance of P1 (2.95). One-quarter would be expected with the e P Var e C = I q σep 2 0 same parental breeds and the same environment for 0 I r σec 2, and purebreds and the crossbreds. The dam variance (0.95) was similar to the sire variance, suggesting that it is mostly genetic rather than environmental. The birth Var (d) =I t σd, 2 litter variances for P1 (0.45) and C1 (0.62) were similar

904 Zumbach et al. Table 3. Estimates of genetic parameters for backfat in the genetic groups 1 (Co)variances Genetic Birth Genetic group 2 Additive Dam litter Residual COV 3 Heritability correlation 4 P1 2.95 0.45 2.97 1.35 0.46 0.83 (0.12) (0.02) (0.07) (0.11) (0.04) (0.09) C1 0.89 0.95 0.62 8.76 0.31 (0.08) (0.08) (0.07) (0.07) (0.02) P2 3.99 0.64 5.86 1.83 0.38 0.89 (0.31) (0.08) (0.18) (0.21) (0.02) (0.05) C2 1.06 0.90 0.80 9.96 0.33 (0.15) (0.12) (0.12) (0.10) (0.02) 1 Standard deviation for (co)variance components; SE for heritability and genetic correlation estimates in parentheses. 2 P1 = purebred Duroc line 1; P2 = purebred Duroc line 2; C1 = P1-sired crossbred; and C2 = P2-sired crossbred. 3 COV = covariance between purebreds and crossbreds. 4 Genetic correlation = genetic correlation between purebreds and crossbreds. in amount and accounted for 5 and 7% of the total variances, respectively. The heritability estimate of the purebred line P1 (0.46) was about 50% greater than that of the crossbreds C1 (0.32). This difference can mainly be explained by the approximately 3-timesgreater residual variance in C1 (8.76). The residual in the model for crossbreds contained the Mendelian sampling. Assuming that the variance of the Mendelian sampling was 2 sire variance, the part of the residual variance in C1 that can be compared with that in P1 was 6.98, which was still more than twice that in P1. The additive genetic variances in P2 (3.99) and C2 (1.06) were greater than in P1 and C1. Also in C2, sire (1.06) and dam (0.90) variances were about 25% of the additive genetic variance in P2. Birth litter variances for P2 (0.64) and C2 (0.80) accounted for 6% of the total variances and were comparable to those in C1 and P1. The heritability estimates for P2 (0.38) and C2 (0.33) were similar, although the residual variance in C2 (9.96) was greater than that in P2 (5.86). The residual variance in P2 was about twice the estimate for P1, which mainly explains the 20% lower heritability. The genetic correlations between purebreds and crossbreds were rather high for both lines (r P1,C1 : 0.83; r P2,C2 : 0.89) in spite of different environments (nucleus and commercial farms) and similar but not identical traits (ultrasound measurement on live pigs and optical measurement on carcasses). This means that the selection for BF in the purebred lines is likely to be effective on the commercial level. Muscle Depth The heritability estimates were lower for MD (Table 4) than for backfat. The additive genetic variances were greatest in the Duroc 1 group, in which the sire variance in C1 (1.61) was about 25% of the additive genetic variance of P1 (6.16). In C1, the dam variance (2.16) was 34% greater than the sire variance, whereas the birth litter variance (0.59) accounted for only 2% of the total variance. In P1, the birth litter variance (0.92) was nearly double that in C1. The residual variance in P1 (13.09) was about double the additive genetic variance. In C1, the residual variance (29.91) was more than double that in P1. As the dam variance in this group was also part of the phenotypic variance, the heritability estimate (0.19) was much lower. The additive genetic variances in P2 (P2: 4.92; C2: 0.98) were lower than those in P1, whereas the other variances were similar for both purebreds and crossbreds. Therefore, heritability estimates of P2 (0.23) and C2 (0.12) were also lower. The sire variance in C2 (0.98) was about one-fifth of the additive genetic variance of P2 (4.92), whereas the dam variance (1.75) was approximately twice the sire variance. However, the genetic variances for sires and dams should be similar. Thus, the dam effect seemed to contain a sizeable environmental effect. As for fat, the genetic correlations between purebreds and crossbreds for MD were also high at around 0.8. Weight at Measurement Variances for crossbreds were adjusted to live BW. The heritability of P1 (0.31) was about twice the value estimated for C1 (0.16; Table 5). The additive genetic sire variance (5.7) was about one-sixth of the additive genetic variance in P1 (34.4). The residual variance in C1 (115.3) was about twice the residual variance of P1 (63.7). The dam variance in C1 (7.0) was at approximately the same level as the sire variance. Birth litter variances in both groups accounted for 11 to 12% of the total variance and were of greater magnitude compared with BF and muscle depth. The genetic correlation between P1 and C1 was 0.53, which is relatively low compared with the correlation of previous traits. This may indicate a larger effect of environmental differences for growth as well as the importance of nonadditive effects.

Correlations between Durocs and crossbreds 905 Table 4. Estimates of genetic parameters for muscle depth in the genetic groups 1 (Co)variances Genetic Birth Genetic group 2 Additive Dam litter Residual COV 3 Heritability correlation 4 P1 6.16 0.92 13.09 2.46 0.31 0.78 (0.31) (0.07) (0.19) (0.23) (0.01) (0.05) C1 1.61 2.16 0.59 29.91 0.19 (0.18) (0.20) (0.20) (0.25) (0.01) P2 4.92 0.73 15.44 1.76 0.23 0.80 (0.48) (0.15) (0.31) (0.27) (0.02) (0.08) C2 0.98 1.75 0.98 30.09 0.12 (0.16) (0.28) (0.29) (0.29) (0.01) 1 Standard deviation for (co)variance components; SE for heritability and genetic correlation estimates in parentheses. 2 P1 = purebred Duroc line 1; P2 = purebred Duroc line 2; C1 = P1-sired crossbred; and C2 = P2-sired crossbred. 3 COV = covariance between purebreds and crossbreds. 4 Genetic correlation = genetic correlation between purebreds and crossbreds. In the Duroc 2 group, the heritability estimates of P2 (0.21) and C2 (0.18) were roughly at the same level. The sire variance in C2 (6.7) was only about 25% of the additive genetic variance in P2 (26.6) but greater than that in C1-P1. The C2 dam variance (4.5) was relatively low, whereas the birth litter variances were similar to those for P1. The residual variance in C2 (117.0) was about 40% greater than that in P2 (84.9). The genetic correlation between C2 and P2 was 0.80, substantially greater than that between P1 and C1. This indicates that there might be a difference in gene frequency between the 2 Duroc populations, implying that nonadditive effects may act in a different way. Purebred selection for weight is likely to be more efficient with P2. WDA The estimates for WDA are provided in Table 6. Variances for crossbreds were adjusted to live BW. The difference between the additive genetic variance of P1 (1178) and the sire variance of C1 (152) was relatively larger than for WGT. However, the heritability estimates were similar (P1: 0.32; C1: 0.16). The proportions of dam (C1: 4.5%) and birth litter variances (P1: 10.9%; C1: 11.5%) were comparable to those for WGT. The genetic correlation between P1 and C1 was 0.60, which was close to the estimate for WGT. The pattern of the estimates for P2-C2 was similar to those for WGT. Weights Another indicator for the efficiency of purebred selection on crossbreds may be the weight for the prediction of breeding values based on purebred information only (Table 7). These weights varied between 0.43 and 0.92 with traits and lines. The greatest values were for BF in both lines (BF1: 0.91; BF2: 0.92), followed by MD (MD1: 0.80; MD2: 0.72). The lowest values were for WGT (0.43) and WDA (0.43) in Duroc line 1. For line Table 5. Estimates of genetic parameters for weight in the genetic groups 1 (Co)variances Genetic Birth Genetic group 2 Additive Dam litter Residual COV 3 Heritability correlation 4 P1 34.4 12.3 63.7 7.4 0.31 0.53 (1.9) (0.5) (1.1) (1.3) (0.01) (0.08) C1 5 5.7 7.0 17.4 115.1 0.16 (0.7) (1.0) (1.1) (0.8) (0.01) P2 26.5 12.6 84.9 10.6 0.21 0.80 (3.0) (1.1) (1.9) (1.8) (0.02) (0.10) C2 5 6.7 4.5 19.0 117.0 0.18 (1.0) (1.4) (1.6) (1.0) (0.01) 1 Standard deviation for (co)variance components; SE for heritability and genetic correlation estimates in parentheses. 2 P1 = purebred Duroc line 1; P2 = purebred Duroc line 2; C1 = P1-sired crossbred; and C2 = P2-sired crossbred. 3 COV = covariance between purebreds and crossbreds. 4 Genetic correlation = genetic correlation between purebreds and crossbreds. 5 (Co)variances and SD adjusted to live BW.

906 Zumbach et al. Table 6. Estimates of genetic parameters for weight per day of age in the genetic groups 1 (Co)variances Genetic Birth Genetic group 2 Additive Dam litter Residual COV 3 Heritability correlation 4 P1 1,178 406 2,139 252 0.32 0.60 (65) (17) (36) (38) (0.01) (0.07) C1 5 152 167 428 2,965 0.16 (18) (24) (26) (20) (0.01) P2 952 411 2,900 332 0.22 0.79 (105) (37) (66) (56) (0.02) (0.07) C2 5 189 119 487 3,087 0.19 (25) (37) (41) (25) (0.01) 1 Standard deviation for (co)variance components; SE for heritability and genetic correlation estimates in parentheses. 2 P1 = purebred Duroc line 1; P2 = purebred Duroc line 2; C1 = P1-sired crossbred; and C2 = P2-sired crossbred. 3 COV = covariance between purebreds and crossbreds. 4 Genetic correlation = genetic correlation between purebreds and crossbreds. 5 (Co)variances and SD adjusted to live BW. 2, the weights were somewhat greater (WGT2: 0.80; WDA2: 0.70). The deviations from the genetic correlations were due to the greater (WGT, WDA) or lower (BF) additive genetic variance in the purebreds compared with the crossbreds. Variance Components DISCUSSION Heritability estimates presented here agree with literature estimates for similar traits (Newcom et al., 2005; van Wijk et al., 2005). Birth litter variances were proportionately smaller for BF and MD compared with those for WGT and WDA, in which preweaning growth is part of the trait. In BF, these litter environmental effects seemed to be of greater importance than in MD. Because dam sire interactions were included in the birth litter variance, this variance also represented the upper bound of the parental dominance variance, which was 25% of the dominance variance. Although P1 and P2 were the same breed and shared the same environmental conditions, their variances differed; P2 generally showed 18 to 97% greater residual Table 7. Weights for the prediction of crossbred breeding values based on purebred information only Trait 1 Weight Backfat 1 0.91 Backfat 2 0.92 Muscle depth 1 0.80 Muscle depth 2 0.72 Pig weight 1 2 0.43 Pig weight 2 2 0.80 Weight per day of age 1 2 0.43 Weight per day of age 2 2 0.70 1 1 = Duroc line 1; 2 = Duroc line 2. 2 Weight adjusted to pig live BW in crossbreds assuming the SD of the carcass weight = 1.2 SD of the pig live BW. variances and, with the exception of BF, 25 to 30% lower additive variances than did P1, leading to lower heritability estimates. Heritability estimates for commercial crossbred pigs were similar for both lines. However, the estimates were up to 50% lower compared with those for the corresponding purebreds. Other studies have reported lower heritabilities for BF and ADG in field data compared with station data (Bidanel and Ducos, 1996; Wolf et al., 2001; Csato et al., 2002; Peskovicova et al., 2002). Genetic Correlations The strongest genetic correlations between nucleus purebred data and corresponding commercial crossbred data were for BF in both lines (P1-C1:0.83; P2-C2: 0.89), followed by MD (P1-C1: 0.78; P2-C2: 0.80). For WGT and WDA, the purebred-crossbred genetic correlations between P2 and C2 (0.80 and 0.79) were larger than those between P1 and C1 (0.53 and 0.60). These values also included genotype environment interactions, because the Duroc nucleus animals were raised under a management with decreased pathogen loads and more pig space compared with the commercial crossbred pigs. Additionally, the traits measured were not identical. Although live animals were measured in the purebred lines, the traits in the commercial pigs were measured on warm carcasses. Divergent results between station and field data were reported. Under Czech and Slovak conditions, the genetic correlations for BF range from 0.72 to 0.84 (Wolf et al., 2001; Peskovicova et al., 2002), whereas the estimates for both French LW and LR are 0.91 (Bidanel and Ducos, 1996). In Hungary, the genetic correlations vary from 0.12 to 0.64 according to the point of measurement (Csato et al., 2002). Wolf et al. (2001) and Peskovicova et al. (2002) reported correlations of approximately 0.5 for ADG. Genetic correlations among similar traits in the literature do not differ much from unity under ad libitum feeding conditions. For BF, Newcom et al. (2005) reat University of Georgia Libraries 1884 on May 14, 2008.

Correlations between Durocs and crossbreds 907 ported a genetic correlation of 0.98 between live animal BF and carcass backfat. Tholen et al. (1998) presented similar results for daily gain when comparing live with carcass weight. Nguyen and McPhee (2005), however, reported estimates of 0.81 for BF and 0.86 for daily gain for pigs under restricted feeding. In this study, the pigs were raised on a few large farms under an ad libitum feeding regimen. Thus, feeding differences did not affect the correlations. Estimates of genetic correlations between BW at 160 and 190 d of age were 0.983 (Huisman et al., 2002). These ages are comparable to those of the P and C groups. The purebred-crossbred genetic correlations obtained for BF in this study were at the upper end of values found in the literature. These varied from 0.32 to 0.98, according to breed combination (Brandt and Täubert, 1998; Merks and Hanenberg, 1998; Lutaaya et al., 2001). The results for WGT and WDA were at the lower end of literature values for WGT and daily gain. Although Merks and Hanenberg (1998) reported only high estimates of 0.90 and 1.00 for 3 different breeds involved, the results of Brandt and Täubert (1998), Lutaaya et al. (2001), and Fischer et al. (2002) were more divergent. They ranged from 0.47 to 0.99, involving 2 to 3 breeds. Although it is better to utilize both purebred and crossbred information unless the test capacity is limited, the importance of crossbred information increases with decreasing purebred-crossbred genetic correlation (Wei and van der Werf, 1994; Bijma and van Arendonk, 1998). The simulations of Bijma and van Arendonk (1998) illustrate that the potential benefit with r pc > 0.9 is small, and a large amount of crossbred information is needed to obtain an additional response. In this study, both the genetic correlations and the weights for the prediction of crossbred breeding values based on purebred information only would justify purebred selection for BF in both populations. For MD, the correlated response would be reduced by 20 (group 1) to 30% (group 2). Although the genetic correlations for WGT and WDA in the Duroc 2 group are reasonably high, the lower weights could justify the inclusion of the crossbred information. The lower r pc (0.5 to 0.6) and weights (0.36) of WGT and WDA found in the Duroc 1 group suggest the use of the crossbred data for an efficient selection in this trait. Lower correlations in WGT and WDA compared with the P2 line indicated a greater plasticity of P1. According to Schinckel et al. (1999), some genetic populations selected for leanness and reduced feed intakes are more sensitive to environmental stressors than some greater feed intake U.S. genetic populations. Environmental sensitivity, also called phenotypic plasticity, is a heritable, evolvable trait (de Jong and Bijma, 2002; Scheiner, 2002). Canalizing selection, i.e., the aptitude to maintain a constant phenotype in fluctuating environments, could also be exploited in animal breeding. SanCristobal-Gaudi et al. (1998) developed indices and approximate expressions of parent-offspring regres- sions for canalizing populations toward an economic optimum. Falconer (1990) showed that stabilizing selection is antagonistic selection in both directions (i.e., selection upwards in a bad environment and downwards in a good environment at the same time) and so is expected to decrease environmental variance. Thus, selection of sires based on crossbred performance, i.e., upward selection in the commercial environment, would be expected to reduce the plasticity in P1 for WGT and WDA. Because plasticity for important traits may be unfavorably correlated, low plasticity for the breeding goal should be desirable (Strandberg, 2005), provided there are high performances in good environments. From an economic point of view, the inclusion of crossbred information is not necessarily suggested. The simulation results of Mielenz et al. (2003), excluding the existence of overdominance, indicate only small extra benefits for high to moderate purebred-crossbred genetic correlation levels when including crossbred information. In addition, it is important to update the genetic parameters frequently for long-term selection (Wei and van der Werf, 1994). There are also alternative indicators for the usefulness of the inclusion of crossbred information, such as the ratio between dominance variance and total genetic variance (Uimari and Gibson, 1998). However, the estimation of dominance variances in most cases is inaccurate due to insufficient sample sizes. Necessary sample sizes for hierarchical full- and half-sib structure were estimated by Mielenz and Schüler (2004). In the case of many sire breeds, genetic evaluation including heterotic effects may be appropriate (Wolf et al., 2006). This study looked at only a few traits and 1 paternal breed. A more complete comprehension concerning the efficiency of purebred selection on commercial animals would also involve maternal breeds, more traits, and possibly more environments. With crosses more complicated than F 1, accurate modeling may be a challenge (Lo et al., 1997). In conclusion, the 2 Duroc lines differed consistently concerning heritability and genetic correlations between purebreds and their respective commercial crossbreds. Heritability estimates were consistently larger for traits measured in P1 and P2 compared with those for traits measured in C1 and C2. Differences in heritabilites between P1 and C1 were consistently larger compared with differences between P2 and C2. Genetic correlations ranged from 0.53 to 0.89. They were consistently lower for traits measured in P1 and C1 compared with traits measured in P2 and C2. The range of correlations between P1 and C1 was larger (0.53 to 0.83) compared with the range of correlations between P2 and C2 (0.79 to 0.89). In addition, growth traits tended to have lower correlations compared with carcass traits. These differences indicate greater environmental sensitivity (plasticity) of P1, which is a leaner line, especially in growth traits. When nucleus and commercial environments differ substantially concerning pathogen

908 Zumbach et al. load, pig space, and other factors, selection strategies should include crossbred data from typical production environments, especially for growth traits. LITERATURE CITED Bidanel, J. P., and A. Ducos. 1996. Genetic correlations between test station and on-farm performance traits in Large White and French Landrace pig breeds. Livest. Prod. Sci. 45:55 62. Bijma, P., and J. A. van Arendonk. 1998. Maximizing genetic gain for the sire line of a crossbreeding scheme utilizing both purebred and crossbred information. Anim. Sci. 66:529 542. Boggs, D. L., R. A. Merkel, M. E. Doumit, and K. W. Bruns. 2006. Livestock and Carcasses. An Integrated Approach to Evaluation, Grading and Selection. Kendall/Hunt Publishing Company, Dubuque, IA. Brandt, H., and H. Täubert. 1998. Parameter estimates for purebred and crossbred performances in pigs. J. Anim. Breed. Genet. 115:97 104. Csato, L., I. Nagy, J. Farkas, and L. Radnoczi. 2002. Genetic parameters of production traits of Hungarian pig populations evaluated in separate and joint (field and station) tests. Arch. Tierz. Dummerstorf 45:375 386. de Jong, G., and P. Bijma. 2002. Selection and phenotypic plasticity in evolutionary biology and animal breeding. Livest. Prod. Sci. 78:195 214. Falconer, D. S. 1990. Selection in different environments: Effects on environmental sensivity (reaction norm) and on mean performance. Genet. Res. Camb. 56:57 70. Fischer, R., U. Müller, and U. Bergfeld. 2002. Genetic association for daily gain (lifetime) in a crossbreeding program. Arch. Tierz. Dummerstorf 45:481 490. Huisman, A. E., R. F. Veerkamp, and J. A. van Arendonk. 2002. Genetic parameters for various random regression models to describe the weight data of pigs. J. Anim. Sci. 80:575 582. Johnson, R. K., E. P. Berg, R. Goodwin, J. W. Mabry, R. K. Miller, O. W. Robison, H. Sellers, and M. D. Tokach. 2004. Evaluation of procedures to predict fat-free lean in swine carcasses. J. Anim. Sci. 82:2428 2441. Lo, L. L., L. Fernando, and M. Grossman. 1997. Genetic evaluation by BLUP in two-breed terminal crossbreeding systems under dominance. J. Anim. Sci. 75:2877 2884. Lutaaya, E., I. Misztal, J. W. Mabry, T. Short, H. H. Timm, and R. Holzbauer. 2001. Genetic parameter estimates from joint evaluation of purebreds and crossbreds in swine using the crossbred model. J. Anim. Sci. 79:3002 3007. Merks, J. W. M., and E. H. A. T. Hanenberg. 1998. Optimal selection strategy for crossbred performance in commercial pig breeding programmes. Proc. 6th World Congr. Genet. Appl. Livest. Prod., Armidale, Australia. 23:575 578. Merks, J., E. Knol, and E. Hanenberg. 2005. Developments in international pig breeding programmes. Proc. 56th Annu. Meet. EAAP, Uppsala, Sweden. Wageningen Academic Publishers, Wageningen, the Netherlands. Mielenz, N., and L. Schüler. 2004. Sample size for estimation of dominance variance with hierarchical full- and half-sib structure. Arch. Tierz. Dummerstorf 47:491 499. Mielenz, N., M. Wensch-Dorendorf, and L. Schüler. 2003. Optimization of the population structure for a two-line crossbreeding scheme a Monte Carlo study. Arch. Tierz. Dummerstorf 46:293 303. Misztal, I., S. Tsuruta, T. Strabel, B. Auvray, T. Druet, and D. Lee. 2002. BLUPF90 and related programs (BGF90). Proc. 7th World Congr. Genet. Appl. Livest. Prod., Montpellier, France. Communication No. 28-07. Newcom, D. W., T. J. Baas, C. R. Schwab, and K. J. Stalder. 2005. Genetic and phenotypic relationships between individual subcutaneous backfat layers and percentage of longissimus intramuscular fat in Duroc swine. J. Anim. Sci. 83:316 323. Nguyen, N. H., and C. P. McPhee. 2005. Genetic parameters and responses of performance and body composition traits in pigs selected for high and low growth rate on a fixed ration over a set time. Genet. Sel. Evol. 37:199 213. Peskovicova, D., J. Wolf, E. Groeneveld, and M. Wolfova. 2002. Simultaneous estimation of the covariance structure of field test, station test and litter recording in pigs. Livest. Prod. Sci. 77:155 165. SanCristobal-Gaudy, M., L. Bodin, J.-M. Elsen, and C. Chevalet. 1998. Prediction of the response to a selection for canalization of a continuous trait in animal breeding. Genet. Sel. Evol. 30:423 451. Scheiner, S. M. 2002. Selection experiments and the study of phenotypic plasticity. J. Evol. Biol. 15:889 898. Schinckel, A. P., B. T. Richert, J. W. Frank, and D. C. Kendall. 1999. Genetic by environmental interactions for pig growth. Pages 93 104 in Proc. Swine Day Rep., West Lafayette, IN. Purdue Univ., West Lafayette, IN. Schinckel, A. P., J. R. Wagner, J. C. Forrest, and M. E. Einstein. 2001. Evaluation of alternative measures of pork carcass composition. J. Anim. Sci. 79:1093 1119. Strandberg, E. 2005. Environmental sensivity and robustness. Page 77 in Proc. 56th Annu. Meet. EAAP, Uppsala, Sweeden. Wageningen Academic Publishers, Wageningen, the Netherlands. Tholen, E., B. Kirstgen, W. Trappmann, and K. Schellander. 1998. Genotype environmental interactions in a German pig breeding herdbook society using crossbred progeny information. Arch. Tierz. Dummerstorf 41:53 63. Uimari, P., and J. P. Gibson. 1998. The value of crossbreeding information in selection of poultry under a dominance model. Anim. Sci. 66:519 528. van Wijk, H. J., D. J. Arts, J. O. Matthews, M. Webster, B. J. Ducro, and E. F. Knol. 2005. Genetic parameters for carcass composition and pork quality estimated in a commercial production chain. J. Anim. Sci. 83:324 333. Via, S., R. Gomulkiewicz, G. de Jong, S. M. Scheiner, C. D. Schlichting, and P. H. van Tienderen. 1995. Adaptive phenotypic plasticity: Consensus and controversy. Trends Ecol. Evol. 10:212 217. Wei, M., and H. A. van der Steen. 1991. Comparison of reciprocal recurrent selection with pure-line selection systems in animal breeding (A review). ABA J. 59:281 298. Wei, M., and J. H. van der Werf. 1994. Maximizing genetic response in crossbreds using both purebred and crossbred information. Anim. Prod. 59:401 413. Wei, M., J. H. van der Werf, and E. W. Brascamp. 1991. Relationship between purebred and crossbred parameters. II. Genetic correlation between purebred and crossbred performance under the model with two loci. J. Anim. Breed. Genet. 108:262 269. Wolf, J., D. Peskovicova, and E. Groeneveld. 2001. Stability of genetic parameter estimates for production traits in pigs. J. Anim. Breed. Genet. 118:161 172. Wolf, J., D. Peskovicova, E. Zakova, and E. Groeneveld. 2006. Additive and heterotic breed effects in the genetic evaluation of pig sire breeds. Anim. Sci. 82:455 462. Wolter, B. F., M. Ellis, J. M. DeDecker, S. E. Curtis, G. R. Hollis, R. D. Shanks, E. N. Parr, and D. M. Webel. 2002. Effects of double stocking and weighing frequency on pig performance in weanto-finish production systems. J. Anim. Sci. 80:1442 1450.

References This article cites 9 articles, 8 of which you can access for free at: http://jas.fass.org/cgi/content/full/85/4/901#bibl