- Email: [email protected]
Correlated responses to divergent selection for phytate phosphorus bioavailability in a randombred chicken population1
Correlated Responses to Divergent Selection for Phytate Phosphorus Bioavailability in a Randombred Chicken Population1 W. Zhang, S. E. Aggrey,2 G. M. Pesti, R. I. Bakalli, and H. M. Edwards, Jr. University of Georgia, Department of Poultry Science, Poultry Science Building, Athens, Georgia 30602-2772 The second evaluation criterion was the average best linear unbiased prediction estimated breeding value (EBV). The results showed asymmetric genetic trends in BW, BWG, and FC, and the correlated responses were mainly due to the genetic changes that occurred in H line because little genetic change occurred in L line across generations. At G3, the line differences of EBV were close to the CRC values for all the traits except FCR. This suggested that CRC and EBV criteria would tend to be consistent with the increase across generations. However, at G1 and G2, the line differences of the EBV actually deviated from the CRC values for BWG and FC. The inconsistency could be attributed to experimental errors and genetic drift that were not accounted by the fixed model for obtaining CRC.
(Key words: correlated response, divergent selection, genetic trend, phenotypic trend, phytate phosphorus bioavailability) 2005 Poultry Science 84:536–542
Phytate phosphorus utilization is a trait of economic importance in poultry because of its obvious nutrition cost and environmental implications (Ravindran et al., 1995). Poultry diets are made primarily of ingredients of plant origin, including cereal grains, cereal by-products, and oil seed meals. About 70% of all P in these plant products is present as phytates (phytic acid and its salts) (Ravindran et al., 1995). Poultry have a very limited ability to utilize phytate P due to the lack of adequate levels of endogenous phytase (Heuser et al., 1943). The inability of poultry to hydrolyze phytate P results in a substantial loss of P in the excreta, which creates a significant pollution threat to the environment when manure (containing residual P) is applied to land (Ravindran et al., 1995).
INTRODUCTION A quantitative trait is assumed affected by many genes. As the result of pleiotropy, the same sets of genes can control more than one trait. The strength of the relationship between traits controlled by the same sets of genes can be measured by the genetic correlations among the traits. When selection pressure is applied to a trait, there would be an according genetic change in the correlated trait(s), depending on the selection intensity and the genetic correlations between them. The correlated responses from selection for economically important traits in livestock breeding and experimental study have been frequently investigated in many kinds of farm animals and poultry (Fredeen and Mikami, 1986; Kaplon et al., 1991; Katle and Kolstad, 1991; Kuhlers and Jungst, 1993; Schulman et al., 1994; Tufvesson et al., 1999; Garcia and Baselga, 2002).
Abbreviation Key: BLUP = best linear unbiased prediction; BWG = body weight gain; CRC = cumulative correlated response; E1 = estimation of variance components using data from generations 0 to 3 in H and L lines; E2 = estimation of variance components using data from generations 0, and 1 to 3 in H lines; E3 = estimation of variance components using data from generations 0, and 1 to 3 in L lines; EBV = estimated breeding value; FC = feed conversion; FCR = feed conversion ratio; G = generation; H line = high phytate phosphorus bioavailability line; L line = low phytate phosphorus bioavailability line; PBA = phytate phosphorus bioavailability.
2005 Poultry Science Association, Inc. Received for publication May 12, 2004. Accepted for publication November 29, 2004. 1 Supported by funds from US Poultry and Egg Association and State and Hatch funds allocated to the Georgia Agricultural Experimental Stations of the University of Georgia. 2 To whom correspondence should be addressed: [email protected].
536
Downloaded from http://ps.oxfordjournals.org/ at Georgian Court University on March 10, 2015
ABSTRACT The current study was undertaken to evaluate the correlated responses to 3 generations of divergent selection for phytate phosphorus bioavailability (PBA) in the Athens-Canadian randombred chicken population. The traits studied were BW at 4 wk of age, BW gain (BWG), feed consumption (FC), and feed conversion ratio (FCR) during a period of 3 d. The first evaluation criterion was the cumulated divergent correlated response (CRC), which was calculated as the line difference of the least square means of phenotypic values for each trait at a given generation after adjustment for sex and hatch effects. The results showed a consistent correlated response in BW across generations. The CRC at generation G3 was 26.8 g (P < 0.01). The chickens in the low PBA line (L line) had higher BW than the high PBA line (H line). The CRC for BWG, FC, and FCR were significant (P < 0.05) only at G3.
537
CORRELATED RESPONSES TO SELECTION FOR PHYTATE PHOSPHORUS
TABLE 1. Descriptive statistics (mean ± SD) of BW at 4 wk, BW gain (BWG), feed consumption (FC), and feed conversion ratio (FCR) in a 3-generation divergent selection program for phytate phosphorus bioavailability (PBA) BW (g) 1
Generation/line G0 G1 G1 G2 G2 G3 G3
(n = 894) − H (n = 359) − L (n = 356) − H (n = 327) − L (n = 343) − H (n = 368) − L (n = 345)
Male 307.87 319.64 323.50 305.15 312.53 317.02 343.29
± ± ± ± ± ± ±
43.54 41.61 40.99 50.88 40.52 49.12 49.23
BWG (g) Female
273.19 283.21 292.45 269.92 276.76 268.83 296.16
± ± ± ± ± ± ±
37.40 30.04 37.14 41.65 38.42 48.03 44.52
Male 47.69 57.47 56.54 53.33 52.65 52.06 56.00
± ± ± ± ± ± ±
11.99 14.52 14.77 10.42 8.32 10.69 9.76
FC (g) Female
41.57 49.36 50.91 43.77 43.82 44.62 47.26
± ± ± ± ± ± ±
11.62 12.72 12.76 9.17 9.22 10.69 10.78
Male 107.51 115.59 113.53 123.56 122.07 124.57 129.30
± ± ± ± ± ± ±
FCR (g:g) Female
16.88 16.11 16.62 18.86 15.01 18.91 17.54
95.25 102.49 102.09 106.73 106.43 108.25 112.13
± ± ± ± ± ± ±
Male
16.66 13.39 14.13 14.43 15.05 20.03 18.22
2.33 2.10 2.09 2.36 2.35 2.45 2.35
± ± ± ± ± ± ±
Female
0.42 0.42 0.41 0.35 0.28 0.41 0.35
2.38 2.17 2.09 2.50 2.48 2.50 2.44
± ± ± ± ± ± ±
0.46 0.46 0.43 0.39 0.34 0.48 0.35
1
G0 to G3 = generation 0 to 3; Line H = high PBA line; Line L = low PBA line.
MATERIALS AND METHODS Birds and Selection Thirty-five sires and 105 dams randomly selected from the Athens-Canadian randombred population were used to generate the base population [generation (G) 0] and each sire was randomly mated with 3 females through artificial insemination. The PBA, and growth and feed utilization performance characteristics of the base population have been reported (Zhang et al., 2003). Birds were ranked according to their hatch-corrected PBA values to establish the divergent sub-populations. Selection. The divergent selection was conducted with 12 males and 36 females in each line at each generation. Details of the divergent selection and the experimental procedures have been described by Zhang et al. (2005). The correlated traits measured were BW, BWG, FC, and FCR. The correlated traits have been described by Zhang et al. (2003).
Yijkl = Li + Sj + Hk + LSij + eijkl where Li = line effect (i = 1, 2), Sj = sex effect (j = 1, 2), Hk = hatch group effect (k = 1,2.....6), and LSij = the interaction of line and sex. Yijkl = Individual observation for a trait and eijkl = residual. For G0, terms Li and were not included. Records with student residual greater than 3 standard deviations for PBA and the correlated traits were considered outliers and were removed. After data editing, 894 records in G0, 715 records in G1, 670 records in G2, and 713 records with complete information for BW, BWG, FC, and FCR in G3 were used for obtaining descriptive statistics (SAS Institute, 1998). The pooled data set was used in the estimation of breeding values (EBV) with mixed model techniques. In addition, 97 parents and 105 grandparents of G0 were included in the analysis. Cumulated divergent correlated responses were calculated as the differences of the least square means of the phenotypic values of traits between the H line and L line at a given generation after adjustment for sex and hatch effects. Genetic trends were calculated by regressing average EBV on generation for each line (Schulman et al., 1994; Sewalem et al., 1998; Bovenhuis et al., 2002; Garcia and Baselga, 2002). A multivariate mixed linear model (Henderson, 1973, 1984; Mrode,1996) was fitted to obtain the best linear unbiased prediction (BLUP) of additive genetic effects of individuals of all generations for BW, BWG, FC, and FCR. Corresponding to each generation and each line, the means of the BLUP value was obtained as average EBV. TABLE 2. Cumulated divergent correlated response (CRC)1 (± SE) in BW, BW gain (BWG), feed consumption (FC), and feed conversion ratio (FCR) to the divergent selection for phytate phosphorus bioavailability (PBA) Trait
Generation 1
Generation 2
Statistical Analysis
BW (g) BWG (g) FC (g) FCR (g:g)
−5.76 −0.07 1.52 0.04
−7.37 0.25 0.71 0.02
The PROC GLM (SAS Institute, 1998) was used to test the effects of sex, hatch group, and line (for G1 to G3), find the outliers in the data set for each generation, and evaluate the CRC. The full model was:
1 CRC was calculated as the difference in phenotypic values for each trait between the high and low PBA lines after adjustment for sex and hatch effects. *P < 0.05. **P < 0.01.
± ± ± ±
2.87* 0.99 1.05 0.03
± ± ± ±
3.32* 0.71 1.17 0.03
Generation 3 −26.80 −3.40 −4.98 0.08
± ± ± ±
3.49** 0.75* 1.24* 0.03*
Downloaded from http://ps.oxfordjournals.org/ at Georgian Court University on March 10, 2015
Genetic selection is a potential approach to the improvement of phytate phosphorus utilization in poultry (Aggrey et al., 2002; Zhang et al., 2003). The genetic basis for the trait has been investigated (Zhang et al., 2003). A negative relationship between phytate phosphorus bioavailability (PBA) and growth rate was implied by Edwards (1983) and verified in a recent study (Zhang, et al., 2003). A modest but significant response was observed from a short-term divergent selection for PBA (Zhang et al., 2005). In the current study, the cumulated divergent correlated responses (CRC) for BW, BW gain (BWG), feed consumption (FC), and feed conversion ratio (FCR) to the divergent selection for PBA and the genetic trends of these traits were investigated. The genetic correlations between PBA and the correlated traits were estimated with several pooled data sets to infer the effect of selection on the genetic covariance.
538
ZHANG ET AL.
Bivariate mixed models were fitted to estimate the genetic correlations of the correlated traits with PBA and a univariate mixed model was used to estimate the heritability for the base population. Three pooled data sets were used in the analysis, which included: 1) whole pooled data across generations and lines (E1), 2) the data from G0 and G1 to G3 in H line (E2), and 3) the data from G0 and G1 to G3 in L line (E3). Genetic parameters were calculated according to the definitions (Falconer and Mackay, 1996). The standard errors were based on the asymptotic variances of the estimate of the cumulative response (Stuard and Ord, 1994; Dodenhoff et al., 1998). The general expression of a mixed model was
y = Xβ + Zu + e where, y = (y1ⴢ y2ⴢ .......ytⴢ)′ and ytⴢ is the vector of observations for trait t; X = matrix that relates fixed effects to records; Z = matrix that relates animals to the records; β = (β1ⴢ ,β2ⴢ .......βtⴢ)ⴢ and βtⴢ is the vector of fixed effects for trait t; u = (u1ⴢ u2ⴢ .......utⴢ)′ and utⴢ is vector of random animal effects (or additive genetic effects) for trait t; and e = (e1ⴢ e2ⴢ .......etⴢ)′ and etⴢ is the vector of residual effects for trait t. The variances of random effects were var(u) = A ⊗ G and var(e) = I ⊗ R, where A is additive relationship matrix, G is the (co)variance matrix for genetic effects of
TABLE 3. Best linear unbiased prediction estimated breeding values (Mean ± SD) for BW at 4 wk, and BW gain (BWG), feed consumption (FC), and feed conversion ratio (FCR) across generations in the divergently selected lines for phytate phosphorus bioavailability (PBA) Generation/line1 G0 G1 G1 G2 G2 G3 G3
(n = 894) − H (n = 359) − L (n = 356) − H (n = 327) − L (n = 343) − H (n = 368) − L (n = 345)
1
BW (g) −1.32 −4.33 2.20 −11.60 −1.72 −25.56 −1.70
± ± ± ± ± ± ±
23.94 25.36 23.75 27.50 27.70 27.39 25.33
BWG (g) −0.58 −0.23 −0.30 −0.88 −1.14 −3.04 −0.99
± ± ± ± ± ± ±
3.82 4.07 4.23 4.16 3.56 4.50 3.87
FC (g) −0.84 0.68 0.46 −2.33 −0.98 −6.54 −0.69
G0 to G3 = generation 0 to 3; Line H = high PBA line; Line L = low PBA line.
± ± ± ± ± ± ±
6.77 6.91 6.64 7.57 6.56 8.07 8.02
FCR (g:g) 0.01 0.00 0.03 0.01 0.03 0.02 0.02
± ± ± ± ± ± ±
0.08 0.06 0.07 0.08 0.06 0.08 0.05
Downloaded from http://ps.oxfordjournals.org/ at Georgian Court University on March 10, 2015
FIGURE 1. Phenotypic trends for BW, body weight gain (BWG), feed consumption (FC), and feed conversion ratio (FCR) in chicken populations selected for high (H line) or low (L line) phytate phosphorus bioavailability.
539
CORRELATED RESPONSES TO SELECTION FOR PHYTATE PHOSPHORUS TABLE 4. Genetic (rA) and residual correlations (rE) (± SE) of phytate phosphorus bioavailability (PBA) with BW at 4 wk, and BW gain (BWG), feed consumption (FC), and feed conversion ratio (FCR) estimated using different pooled data sets1 Estimation 1 Trait BW BWG FC FCR
rA −0.21 0.00 −0.18 −0.45
± ± ± ±
Estimation 2 rE
0.01 0.12 0.12 0.14
−0.12 0.08 0.08 −0.13
± ± ± ±
rA 0.03 0.02 0.02 0.01
−0.30 0.03 −0.35 −0.46
± ± ± ±
Estimation 3 rE
0.12 0.15 0.13 0.15
0.07 0.10 −0.02 −0.12
± ± ± ±
rA 0.04 0.01 0.02 0.01
−0.06 0.24 −0.10 −0.35
± ± ± ±
rE 0.12 0.14 0.13 0.14
−0.15 0.04 −0.05 −0.11
± ± ± ±
0.05 0.03 0.03 0.02
1 Estimation 1 = estimated using the whole pooled data; estimation 2 = data from generation 0 and generations 1 to 3 of the high PBA line; estimation 3 = data from generation 0 and generations 1 to 3 of the low PBA line.
储 θˆ t − θˆ t−1 储2/储 θˆ t 储2 < 5 × 10−11 where θˆ t is the vector of estimated parameters in the tth iteration, and the Delta convergence was lower than 5 × 10−6.
RESULTS AND DISCUSSION Cumulated Divergent Correlated Responses The descriptive statistics of BW, BWG, FC, and FCR in generations 1 to 3 from the fixed-effect model are listed in Table 1. The phenotypic trends in BW, BWG, FC, and FCR to the divergent selection for PBA are shown in Figure 1. In general, the means of the traits fluctuated over generations. The comparisons (Table 2) between lines demonstrated a small CRC (P < 0.05) in BW (−5.8 g) after one generation of selection (at G1), and the L line had higher BW than the H line. After 3 generations of selection, the CRC in BW had reached −26.8 g (P < 0.01). The CRC in BWG, FC, and FCR were not significant at G1 and G2 (P > 0.05). However, at G3 there were significant CRC for BWG (P < 0.01), FC (P < 0.01), and FCR (P < 0.05). The H line had lower BW, BWG, FC, and higher FCR, compared with the L line. The CRC showed a trend across the generations in FC. However, the trends in BWG and FCR were not clear.
Divergence of EBV and Genetic Trends The descriptive statistics of BLUP EBV for BW, BWG, FC, and FCR at G0 to G3 for the H and L lines are summa-
rized in Table 3. The differences of EBV for these traits between the H line and the L line were similar to the phenotypic responses. The line differences in EBV for BW increased across the generations. However, the line difference for FC only occurred at G2 and G3, and for BWG only at G3. For FCR, there was a small line difference in EBV at G1 and G2, and the difference was nearly zero at G3. The EBV for all traits (Table 3, Figure 2) did not fluctuate as their phenotypic values in both H and L lines. The EBV for BW, BWG, and FC genetically tended to decrease in the H line and almost remained unchanged in the L line. It seems that the differences between lines for these traits were mainly due to the changes of EBV that occurred in the H line. The regression coefficients of the average EBV on the generations were −8.01, −0.80, −2.01, and 0.00, for BW, BWG, FC, and FCR, respectively, in the H line. Among all the traits, only BW in the H line had a slope (P < 0.05). Similarly, the regression coefficients were −0.506, −0.136, −0.096, and 0.00 for BW, BWG, FC, and FCR, respectively, in the L line. None of the regression coefficients in the L line was different from zero (P > 0.05).
Estimation of Genetic Parameters of the Related Traits Using Pooled Data Sets The genetic correlations (rA) between PBA and BW, BWG, FC, and FCR are presented in Table 4 and the heritabilities (h2) for BW, BWG, FC, and FCR estimated with different pooled data sets are listed in Table 5. The heritabilities of BW, FC, and FCR were consistent in E1 to E3. The negative genetic correlations of PBA with BW, and FC were stronger in E2 than in E3, and the values in E1 were between those of E2 and E3. The negative genetic correlations between FCR and PBA were high in all 3 estimations. The genetic correlation between PBA and BWG was positive [0.24 ± 0.14 (SE)] in E3, however, this correlation was zero and very low in E1 and E2, respectively. The phenotypic data exhibited fluctuations across generations and still had not formed a trend after selecting for 3 generations. This could be attributed to environmental factors, such as variation in feed ingredients, excreta grinding, and temperature. It also means that, in a shortterm selection project when genetic progress is relatively small compared with the environmental effects, we can-
Downloaded from http://ps.oxfordjournals.org/ at Georgian Court University on March 10, 2015
the studied traits, I is an identity matrix, and R is the (co)variance matrix for the corresponding residual effects. Generations and hatch groups within generations were combined into 24 generation-hatch groups. Sex and generation-hatch groups were included in the model as fixed effects. The formation of A−1 was based on Henderson’s (1975) and Quaas’s (1976) methods. The calculation of BLUP and variance components was accomplished using the average information algorithm for restricted maximum likelihood (REML) procedure (Johnson and Thompson, 1995) with an AiREML program (Misztal et al., 2002). Convergence was considered to have been reached when
540
ZHANG ET AL.
not expect to infer from the phenotypic trend. Evaluation of the correlated responses should therefore be determined from both line comparison, and the estimation of breeding values. When a comparison of the breeding values at G3 was made, the differences of the EBV between H line and L line were close to the CRC values for BW, BWG, and FC. They were −23.86 vs. −26.80 g for BW; −2.05 vs. −3.40 g for BWG; and −5.85 vs. −4.98 g for FC. However, at G1 and G2, the line differences of EBV deviated from the CRC values for BWG and FC. This inconsistency could be attributed to experimental errors and genetic drift that were not accounted for by the fixed model for obtaining CRC. This also means the least-square analysis based on line comparisons sometimes did not indicate the true genetic changes that had occurred in a short-term selection project (Sorensen and Kennedy, 1984; Kennedy, 1990; Walsh, 2004). A trend in FCR for either CRC or EBV criteria could not be ascertained. This implies that in the short-term, selection for PBA did not affect feed efficiency. This result was confirmed by the small genetic correlations between PBA and FCR estimated with the records of the base
population. Theoretically, simple linear equations can be established by regressing EBV on generations for predicting the genetic trends of the related traits (Falconer and Mackey, 1996; Walsh, 2004), but in the current study, selection was undertaken only for 3 generations, and the genetic changes were relatively small and not stable for most traits, and consequently only the equation for BW in the H line had a slope coefficient (P < 0.05). Therefore, data from more generations of selection will be needed for establishing useful prediction equations for BWG and FCR. The profile of the genetic correlations of PBA and the correlated traits estimated using the pooled data set did not follow any pattern. In the estimation using the data of the base population exclusively, the genetic correlations of PBA with BW, BWG, and FC were negative and moderate (Zhang et al., 2003). This was in agreement with genetic trends in both the H and L lines. However, when information of the selected generations (G1 to G3) was combined, the relationship tended to become weaker, even reversed in direction, and showed line dimorphism (Table 4).
Downloaded from http://ps.oxfordjournals.org/ at Georgian Court University on March 10, 2015
FIGURE 2. Genetic trends for BW, body weight gain (BWG), feed consumption (FC), and feed conversion ratio (FCR) in chicken populations selected for high (H line) or low (L line) phytate phosphorus bioavailability.
541
CORRELATED RESPONSES TO SELECTION FOR PHYTATE PHOSPHORUS TABLE 5. Genetic variance (σ2A) and heritability (h2) (± SE) of BW at 4 wk, BW gain (BWG), feed consumption (FC), and feed conversion ratio (FCR) estimated using different pooled data sets Estimation 1 Trait BW BWG FC FCR
σ2A 1,080.89 32.35 92.75 0.0
± ± ± ±
Estimation 2 h2
78.62 8.12 13.44 1.00
0.58 0.27 0.41 0.08
± ± ± ±
σ2A 0.03 0.06 0.05 0.02
1,114.03 31.61 85.40 0.02
± ± ± ±
Estimation 3 h2
84.27 7.68 13.68 0.00
0.60 0.26 0.38 0.09
± ± ± ±
σ2A 0.03 0.06 0.05 0.01
1,003.51 30.41 94.02 0.01
± ± ± ±
h2 94.82 10.16 16.21 0.00
0.57 0.26 0.43 0.09
± ± ± ±
0.04 0.08 0.06 0.02
1 Estimation 1 = estimated using the whole pooled data; estimation 2 = data from generation 0 and generations 1 to 3 of the high PBA line; estimation 3 = data from generation 0 and generations 1 to 3 of the low PBA line.
contributing factors for the moderately negative genetic correlations of PBA with BW, BWG, and FC, which were estimated using the data of the base population exclusively (Zhang et al., 2003). Selection for PBA could change the frequency of the allele(s) disproportionately in the divergent lines leading to the asymmetric responses in the correlated traits. It is also possible that genetic homeostasis between PBA and the correlated traits were different in the H and L lines. In summary, from the current investigation it can be inferred that direct selection for PBA can consequently led to concurrent changes in BW, BWG, FC, and FCR in the H line, however, similar changes in the L line was less appreciable.
ACKNOWLEDGMENTS The authors thank Ignacy Misztal and Shogo Tsuruta, Department of Animal and Dairy Science, University of Georgia (Athens, GA) for their permission to modify and use their program, and Cheryl Pearson Gresham, Mohamet Komberi, Donna Slavin, and Elaine Foster, Department of Poultry Science, University of Georgia, for their technical assistance.
REFERENCES Aggrey, S. E., W. Zhang, R. I. Bakalli, G. M. Pesti, and H. M. Edwards, Jr. 2002. Genetics of phytate phosphorus bio-availability in poultry. Pages 277–279 in Proceedings of the 7th World Congress on Genetics Applied to Livestock Production. Montpellier, France. Bovenhuis, H., H. Bralten, M. G. B. Niuwland, and H. K. Parmentier. 2002. Genetic parameters for antibody response of chickens to sheep red blood cells based on a selection experiment. Poult. Sci. 81:309–315. Dodenhoff, J., L. D. Van Vleck, S. D. Kachman, and R. M. Kock. 1998. Parameter estimates for direct, maternal, and grandmaternal genetic effects for birth weight and weaning weight in Hereford cattle. J. Anim. Sci. 76:2521–2547. Dunnington, E. A., and P. B. Siegel. 1996. Long-term divergent selection for eight-week body weight in White Plymouth Rock chickens. Poult. Sci. 75:1168–1179. Edwards, H. M., Jr. 1983. Phosphorus I. Effect of breed and strain on utilization of sub-optimal level of phosphorus in ration. Poult. Sci. 62:77–84. Falconer, D. S., and T. F. C. Mackay. 1996. Introduction to Quantitative Genetics. Longman Group, Essex, UK. Fredeen, H. T., and H. Mikami. 1986. Mass selection in a pig population: Correlated changes in carcass merit. J. Anim. Sci. 62:1654.
Downloaded from http://ps.oxfordjournals.org/ at Georgian Court University on March 10, 2015
In the estimation (E2) using the combined data of G0 and the H line, the genetic correlation between PBA and BW was moderately negative and this was consistent with the genetic trends in the H line (Table 4). In the estimation (E3) using the combined data of G0 and L line, the genetic correlation between PBA and BW was almost negligible and this was consistent with the lack of genetic changes in L line. A similar phenomenon was observed for FC. The genetic correlation between PBA and BWG in the estimations using pooled data sets (E1 to E3) was unexpected and not in agreement with the genetic trend profile. The selection showed a negative genetic trend in the H line, yet the genetic correlation was negligible in the estimation (E2) using the combined data from G0 and the H line (Table 4). The large negative genetic correlation between PBA and FCR in E1 to E3 is difficult to interpret. It seems that the genetic covariance of PBA and FCR was unstable because their genetic variances were small and their measurement was subject to substantial errors in the short-term selection experiment. The mechanism underlying the inconsistent estimates of genetic correlations using different data sets is still not clear. However, Dunnington and Siegel (1996) suggested that changes in resource allocation, or prioritization of resource use of demands placed on artificially selected chickens may be responsible for changes in the genetic covariance among traits. Resource allocation or prioritization of resource use could be the underlying factors behind the asymmetric correlated responses in BW, BWG, and FC. Another possible explanation for the asymmetric correlated responses could be the involvement of major gene(s). Some major genes have modest effects on the associated metric traits, and do not significantly influence the distribution profiles of the phenotypic and genetic values when the frequencies are low. For example, a recessive and sexlinked dwarfing allele, dwB, in poultry was found to be associated with only a 10% BW reduction, smaller than a standard deviation (Shiva-Prasad and Jaap, 1977). Even though the inheritance of PBA did not deviate from an additive model of many loci (Zhang et al., 2003, 2005), it is still possible that a recessive allele(s) with a “large effect” and a low frequency in the base population in addition to genes of small effect controlled PBA. The pleiotropic effects of the allele(s) may be associated with high PBA and low growth rate. This could be one of the
542
ZHANG ET AL. Ravindran,V., W. L. Bryden, and E.T. Kornegay. 1995. Phytates: Occurrence, bioavailability, and implications in poultry nutrition. Poult. Avian Biol. Rev. 6:125–143. SAS Institute. 1998. SAS version 8.12. SAS Institute Inc., Cary, NC. Sewalem, A., K. Lillpers, and M. Wilhemson. 1998. Genetic trends of production and reproduction traits in White Leghorn lines selected for production traits. Pages 286–289 in Proceedings of the 6th World Congress on Genetics Applied to Livestock Production, Armidale, Australia. Schulman, N., M. Tuiskula-Haavista, L. Siitonen, and E. A. Ma¨ntysaari. 1994. Genetic variation of residual feed consumption in a selected Finnish egg-layer population. Poult. Sci. 73:1479–1484. Shiva-Prasad, H. L., and R. G. Jaap. 1977. Egg and yolk production as influenced by liver weight, liver lipid and plasma lipid in three strains of small bodied chickens. Poult. Sci. 73:1479–1484. Sorensen D. A., and B. W. Kennedy. 1984. Estimation of genetic variances from unselected and selected populations. J. Anim. Sci. 59:1213–1223. Stuard, A., and J. K. Ord. 1994. Kendall’s Advanced Theory of Statistics. Vol. 1. Distribution Theory. 6th ed. Halsted Press, New York, NY. Tufvessen, M., B. Tufvessen, T. von Schantz, K. Johansson, and M. Wilhelmson. 1999. Selection for sexual male characters and their effects on other fitness related traits in White Leghorn chickens. J. Anim. Breed. Genet. 116:127–138. Walsh, B. 2004. Subject: Evolution and Selection of Quantitative Traits. Pages 165–237. http://nitro.biosci.arizona.edu/ zbook/volume_2/vol2.html. Accessed July 1, 2004. Zhang, W., S. E. Aggrey, G. M. Pesti, H. M. Edwards, Jr., and R. I. Bakalli. 2003. Genetics of phytate phosphorus bioavailability: Heritability and genetic correlations with growth and feed utilization traits in a randombred chicken population. Poult. Sci. 82:1975–1979. Zhang, W., S. E. Aggrey, G. M. Pesti, R. I. Bakalli, and H. M. Edwards, Jr. 2005. Genetic analysis on the direct response to divergent selection for phytate phosphorus bioavailability in a randombred chicken population. Poult. Sci. 84:370–375.
Downloaded from http://ps.oxfordjournals.org/ at Georgian Court University on March 10, 2015
Garcia, M. L., and A. Baselga. 2002. Estimation of correlated response on growth traits to selection in litter size of rabbits using a cryopreserved control population and genetic trends. Livest. Prod. Sci. 78:91–98. Henderson, C. R. 1973. Sire evaluation and genetic trends. Pages 10–41 in Proceedings of Animal Breeding and Genetics Symposium in honor of Dr. J. L. Lush. American Society of Animal Science, Champaign, IL. Henderson , C. R. 1975. Rapid method for computing the inverse of a relationship matrix. J. Dairy Sci. 58:1727–1730. Henderson, C. R. 1984. Applications of Linear Models in Animal Breeding. University of Guelph, Ontario, Canada. Heuser, G. F., L. C. Norris, J. McGinnis, and M. L. Scott. 1943. Further evidence of the need for supplementing soybean meal chick rations with phosphorus. Poult. Sci. 22:269–270. Johnson, D. H., and R. Thompson. 1995. Restricted maximum likelihood estimation of variance components for univariate animal models using sparse matrix techniques and average information. J. Dairy Sci. 78:423–447. Kaplon, M. J., M. F. Rothschild, P. J. Berger, and M. Healey. 1991. Genetic and phenotypic trends in Polish Large White nucleus herds. J. Anim. Sci. 69:551–558. Katle, J., and N. Kolstad. 1991. Selection for efficiency of food utilization in laying hens: Direct response in residual food consumption and correlated responses in weight gain, egg production and body weight. Br. Poult. Sci. 32:939–953. Kennedy, B. W. 1990. Mixed model methodology in analysis of designed experiments. Pages 76–94 in Advances in Statistical Methods for Genetic Improvement of Livestock. D. Gianola, and K. Hammond, ed. Springer-Verlag, Berlin, Germany. Kuhlers, D. L., and S. B. Jungst. 1993. Correlated responses in reproductive and carcass traits to selection for 200-day weight in Landrace pigs. J. Anim. Sci. 71:595–601. Misztal, I., S. Tsuruta, T. Strabel, B. Auvray, T. Druet, and D. H. Lee. 2002. BLUPF90 and related programs (BGF90). Pages 743–744 in Proceedings of the 7th World Congress on Genetics Applied to Livestock Production, Montpellier, France. Mrode, R. A. 1996. Linear Models for the Prediction of Animal Breeding Values. CABI, Wallingford, UK. Quaas R. L. 1976. Computing the diagonal elements and the inverse of a large numerator relationship matrix. Biometrica 32:949–953.
(Key words: correlated response, divergent selection, genetic trend, phenotypic trend, phytate phosphorus bioavailability) 2005 Poultry Science 84:536–542
Phytate phosphorus utilization is a trait of economic importance in poultry because of its obvious nutrition cost and environmental implications (Ravindran et al., 1995). Poultry diets are made primarily of ingredients of plant origin, including cereal grains, cereal by-products, and oil seed meals. About 70% of all P in these plant products is present as phytates (phytic acid and its salts) (Ravindran et al., 1995). Poultry have a very limited ability to utilize phytate P due to the lack of adequate levels of endogenous phytase (Heuser et al., 1943). The inability of poultry to hydrolyze phytate P results in a substantial loss of P in the excreta, which creates a significant pollution threat to the environment when manure (containing residual P) is applied to land (Ravindran et al., 1995).
INTRODUCTION A quantitative trait is assumed affected by many genes. As the result of pleiotropy, the same sets of genes can control more than one trait. The strength of the relationship between traits controlled by the same sets of genes can be measured by the genetic correlations among the traits. When selection pressure is applied to a trait, there would be an according genetic change in the correlated trait(s), depending on the selection intensity and the genetic correlations between them. The correlated responses from selection for economically important traits in livestock breeding and experimental study have been frequently investigated in many kinds of farm animals and poultry (Fredeen and Mikami, 1986; Kaplon et al., 1991; Katle and Kolstad, 1991; Kuhlers and Jungst, 1993; Schulman et al., 1994; Tufvesson et al., 1999; Garcia and Baselga, 2002).
Abbreviation Key: BLUP = best linear unbiased prediction; BWG = body weight gain; CRC = cumulative correlated response; E1 = estimation of variance components using data from generations 0 to 3 in H and L lines; E2 = estimation of variance components using data from generations 0, and 1 to 3 in H lines; E3 = estimation of variance components using data from generations 0, and 1 to 3 in L lines; EBV = estimated breeding value; FC = feed conversion; FCR = feed conversion ratio; G = generation; H line = high phytate phosphorus bioavailability line; L line = low phytate phosphorus bioavailability line; PBA = phytate phosphorus bioavailability.
2005 Poultry Science Association, Inc. Received for publication May 12, 2004. Accepted for publication November 29, 2004. 1 Supported by funds from US Poultry and Egg Association and State and Hatch funds allocated to the Georgia Agricultural Experimental Stations of the University of Georgia. 2 To whom correspondence should be addressed: [email protected].
536
Downloaded from http://ps.oxfordjournals.org/ at Georgian Court University on March 10, 2015
ABSTRACT The current study was undertaken to evaluate the correlated responses to 3 generations of divergent selection for phytate phosphorus bioavailability (PBA) in the Athens-Canadian randombred chicken population. The traits studied were BW at 4 wk of age, BW gain (BWG), feed consumption (FC), and feed conversion ratio (FCR) during a period of 3 d. The first evaluation criterion was the cumulated divergent correlated response (CRC), which was calculated as the line difference of the least square means of phenotypic values for each trait at a given generation after adjustment for sex and hatch effects. The results showed a consistent correlated response in BW across generations. The CRC at generation G3 was 26.8 g (P < 0.01). The chickens in the low PBA line (L line) had higher BW than the high PBA line (H line). The CRC for BWG, FC, and FCR were significant (P < 0.05) only at G3.
537
CORRELATED RESPONSES TO SELECTION FOR PHYTATE PHOSPHORUS
TABLE 1. Descriptive statistics (mean ± SD) of BW at 4 wk, BW gain (BWG), feed consumption (FC), and feed conversion ratio (FCR) in a 3-generation divergent selection program for phytate phosphorus bioavailability (PBA) BW (g) 1
Generation/line G0 G1 G1 G2 G2 G3 G3
(n = 894) − H (n = 359) − L (n = 356) − H (n = 327) − L (n = 343) − H (n = 368) − L (n = 345)
Male 307.87 319.64 323.50 305.15 312.53 317.02 343.29
± ± ± ± ± ± ±
43.54 41.61 40.99 50.88 40.52 49.12 49.23
BWG (g) Female
273.19 283.21 292.45 269.92 276.76 268.83 296.16
± ± ± ± ± ± ±
37.40 30.04 37.14 41.65 38.42 48.03 44.52
Male 47.69 57.47 56.54 53.33 52.65 52.06 56.00
± ± ± ± ± ± ±
11.99 14.52 14.77 10.42 8.32 10.69 9.76
FC (g) Female
41.57 49.36 50.91 43.77 43.82 44.62 47.26
± ± ± ± ± ± ±
11.62 12.72 12.76 9.17 9.22 10.69 10.78
Male 107.51 115.59 113.53 123.56 122.07 124.57 129.30
± ± ± ± ± ± ±
FCR (g:g) Female
16.88 16.11 16.62 18.86 15.01 18.91 17.54
95.25 102.49 102.09 106.73 106.43 108.25 112.13
± ± ± ± ± ± ±
Male
16.66 13.39 14.13 14.43 15.05 20.03 18.22
2.33 2.10 2.09 2.36 2.35 2.45 2.35
± ± ± ± ± ± ±
Female
0.42 0.42 0.41 0.35 0.28 0.41 0.35
2.38 2.17 2.09 2.50 2.48 2.50 2.44
± ± ± ± ± ± ±
0.46 0.46 0.43 0.39 0.34 0.48 0.35
1
G0 to G3 = generation 0 to 3; Line H = high PBA line; Line L = low PBA line.
MATERIALS AND METHODS Birds and Selection Thirty-five sires and 105 dams randomly selected from the Athens-Canadian randombred population were used to generate the base population [generation (G) 0] and each sire was randomly mated with 3 females through artificial insemination. The PBA, and growth and feed utilization performance characteristics of the base population have been reported (Zhang et al., 2003). Birds were ranked according to their hatch-corrected PBA values to establish the divergent sub-populations. Selection. The divergent selection was conducted with 12 males and 36 females in each line at each generation. Details of the divergent selection and the experimental procedures have been described by Zhang et al. (2005). The correlated traits measured were BW, BWG, FC, and FCR. The correlated traits have been described by Zhang et al. (2003).
Yijkl = Li + Sj + Hk + LSij + eijkl where Li = line effect (i = 1, 2), Sj = sex effect (j = 1, 2), Hk = hatch group effect (k = 1,2.....6), and LSij = the interaction of line and sex. Yijkl = Individual observation for a trait and eijkl = residual. For G0, terms Li and were not included. Records with student residual greater than 3 standard deviations for PBA and the correlated traits were considered outliers and were removed. After data editing, 894 records in G0, 715 records in G1, 670 records in G2, and 713 records with complete information for BW, BWG, FC, and FCR in G3 were used for obtaining descriptive statistics (SAS Institute, 1998). The pooled data set was used in the estimation of breeding values (EBV) with mixed model techniques. In addition, 97 parents and 105 grandparents of G0 were included in the analysis. Cumulated divergent correlated responses were calculated as the differences of the least square means of the phenotypic values of traits between the H line and L line at a given generation after adjustment for sex and hatch effects. Genetic trends were calculated by regressing average EBV on generation for each line (Schulman et al., 1994; Sewalem et al., 1998; Bovenhuis et al., 2002; Garcia and Baselga, 2002). A multivariate mixed linear model (Henderson, 1973, 1984; Mrode,1996) was fitted to obtain the best linear unbiased prediction (BLUP) of additive genetic effects of individuals of all generations for BW, BWG, FC, and FCR. Corresponding to each generation and each line, the means of the BLUP value was obtained as average EBV. TABLE 2. Cumulated divergent correlated response (CRC)1 (± SE) in BW, BW gain (BWG), feed consumption (FC), and feed conversion ratio (FCR) to the divergent selection for phytate phosphorus bioavailability (PBA) Trait
Generation 1
Generation 2
Statistical Analysis
BW (g) BWG (g) FC (g) FCR (g:g)
−5.76 −0.07 1.52 0.04
−7.37 0.25 0.71 0.02
The PROC GLM (SAS Institute, 1998) was used to test the effects of sex, hatch group, and line (for G1 to G3), find the outliers in the data set for each generation, and evaluate the CRC. The full model was:
1 CRC was calculated as the difference in phenotypic values for each trait between the high and low PBA lines after adjustment for sex and hatch effects. *P < 0.05. **P < 0.01.
± ± ± ±
2.87* 0.99 1.05 0.03
± ± ± ±
3.32* 0.71 1.17 0.03
Generation 3 −26.80 −3.40 −4.98 0.08
± ± ± ±
3.49** 0.75* 1.24* 0.03*
Downloaded from http://ps.oxfordjournals.org/ at Georgian Court University on March 10, 2015
Genetic selection is a potential approach to the improvement of phytate phosphorus utilization in poultry (Aggrey et al., 2002; Zhang et al., 2003). The genetic basis for the trait has been investigated (Zhang et al., 2003). A negative relationship between phytate phosphorus bioavailability (PBA) and growth rate was implied by Edwards (1983) and verified in a recent study (Zhang, et al., 2003). A modest but significant response was observed from a short-term divergent selection for PBA (Zhang et al., 2005). In the current study, the cumulated divergent correlated responses (CRC) for BW, BW gain (BWG), feed consumption (FC), and feed conversion ratio (FCR) to the divergent selection for PBA and the genetic trends of these traits were investigated. The genetic correlations between PBA and the correlated traits were estimated with several pooled data sets to infer the effect of selection on the genetic covariance.
538
ZHANG ET AL.
Bivariate mixed models were fitted to estimate the genetic correlations of the correlated traits with PBA and a univariate mixed model was used to estimate the heritability for the base population. Three pooled data sets were used in the analysis, which included: 1) whole pooled data across generations and lines (E1), 2) the data from G0 and G1 to G3 in H line (E2), and 3) the data from G0 and G1 to G3 in L line (E3). Genetic parameters were calculated according to the definitions (Falconer and Mackay, 1996). The standard errors were based on the asymptotic variances of the estimate of the cumulative response (Stuard and Ord, 1994; Dodenhoff et al., 1998). The general expression of a mixed model was
y = Xβ + Zu + e where, y = (y1ⴢ y2ⴢ .......ytⴢ)′ and ytⴢ is the vector of observations for trait t; X = matrix that relates fixed effects to records; Z = matrix that relates animals to the records; β = (β1ⴢ ,β2ⴢ .......βtⴢ)ⴢ and βtⴢ is the vector of fixed effects for trait t; u = (u1ⴢ u2ⴢ .......utⴢ)′ and utⴢ is vector of random animal effects (or additive genetic effects) for trait t; and e = (e1ⴢ e2ⴢ .......etⴢ)′ and etⴢ is the vector of residual effects for trait t. The variances of random effects were var(u) = A ⊗ G and var(e) = I ⊗ R, where A is additive relationship matrix, G is the (co)variance matrix for genetic effects of
TABLE 3. Best linear unbiased prediction estimated breeding values (Mean ± SD) for BW at 4 wk, and BW gain (BWG), feed consumption (FC), and feed conversion ratio (FCR) across generations in the divergently selected lines for phytate phosphorus bioavailability (PBA) Generation/line1 G0 G1 G1 G2 G2 G3 G3
(n = 894) − H (n = 359) − L (n = 356) − H (n = 327) − L (n = 343) − H (n = 368) − L (n = 345)
1
BW (g) −1.32 −4.33 2.20 −11.60 −1.72 −25.56 −1.70
± ± ± ± ± ± ±
23.94 25.36 23.75 27.50 27.70 27.39 25.33
BWG (g) −0.58 −0.23 −0.30 −0.88 −1.14 −3.04 −0.99
± ± ± ± ± ± ±
3.82 4.07 4.23 4.16 3.56 4.50 3.87
FC (g) −0.84 0.68 0.46 −2.33 −0.98 −6.54 −0.69
G0 to G3 = generation 0 to 3; Line H = high PBA line; Line L = low PBA line.
± ± ± ± ± ± ±
6.77 6.91 6.64 7.57 6.56 8.07 8.02
FCR (g:g) 0.01 0.00 0.03 0.01 0.03 0.02 0.02
± ± ± ± ± ± ±
0.08 0.06 0.07 0.08 0.06 0.08 0.05
Downloaded from http://ps.oxfordjournals.org/ at Georgian Court University on March 10, 2015
FIGURE 1. Phenotypic trends for BW, body weight gain (BWG), feed consumption (FC), and feed conversion ratio (FCR) in chicken populations selected for high (H line) or low (L line) phytate phosphorus bioavailability.
539
CORRELATED RESPONSES TO SELECTION FOR PHYTATE PHOSPHORUS TABLE 4. Genetic (rA) and residual correlations (rE) (± SE) of phytate phosphorus bioavailability (PBA) with BW at 4 wk, and BW gain (BWG), feed consumption (FC), and feed conversion ratio (FCR) estimated using different pooled data sets1 Estimation 1 Trait BW BWG FC FCR
rA −0.21 0.00 −0.18 −0.45
± ± ± ±
Estimation 2 rE
0.01 0.12 0.12 0.14
−0.12 0.08 0.08 −0.13
± ± ± ±
rA 0.03 0.02 0.02 0.01
−0.30 0.03 −0.35 −0.46
± ± ± ±
Estimation 3 rE
0.12 0.15 0.13 0.15
0.07 0.10 −0.02 −0.12
± ± ± ±
rA 0.04 0.01 0.02 0.01
−0.06 0.24 −0.10 −0.35
± ± ± ±
rE 0.12 0.14 0.13 0.14
−0.15 0.04 −0.05 −0.11
± ± ± ±
0.05 0.03 0.03 0.02
1 Estimation 1 = estimated using the whole pooled data; estimation 2 = data from generation 0 and generations 1 to 3 of the high PBA line; estimation 3 = data from generation 0 and generations 1 to 3 of the low PBA line.
储 θˆ t − θˆ t−1 储2/储 θˆ t 储2 < 5 × 10−11 where θˆ t is the vector of estimated parameters in the tth iteration, and the Delta convergence was lower than 5 × 10−6.
RESULTS AND DISCUSSION Cumulated Divergent Correlated Responses The descriptive statistics of BW, BWG, FC, and FCR in generations 1 to 3 from the fixed-effect model are listed in Table 1. The phenotypic trends in BW, BWG, FC, and FCR to the divergent selection for PBA are shown in Figure 1. In general, the means of the traits fluctuated over generations. The comparisons (Table 2) between lines demonstrated a small CRC (P < 0.05) in BW (−5.8 g) after one generation of selection (at G1), and the L line had higher BW than the H line. After 3 generations of selection, the CRC in BW had reached −26.8 g (P < 0.01). The CRC in BWG, FC, and FCR were not significant at G1 and G2 (P > 0.05). However, at G3 there were significant CRC for BWG (P < 0.01), FC (P < 0.01), and FCR (P < 0.05). The H line had lower BW, BWG, FC, and higher FCR, compared with the L line. The CRC showed a trend across the generations in FC. However, the trends in BWG and FCR were not clear.
Divergence of EBV and Genetic Trends The descriptive statistics of BLUP EBV for BW, BWG, FC, and FCR at G0 to G3 for the H and L lines are summa-
rized in Table 3. The differences of EBV for these traits between the H line and the L line were similar to the phenotypic responses. The line differences in EBV for BW increased across the generations. However, the line difference for FC only occurred at G2 and G3, and for BWG only at G3. For FCR, there was a small line difference in EBV at G1 and G2, and the difference was nearly zero at G3. The EBV for all traits (Table 3, Figure 2) did not fluctuate as their phenotypic values in both H and L lines. The EBV for BW, BWG, and FC genetically tended to decrease in the H line and almost remained unchanged in the L line. It seems that the differences between lines for these traits were mainly due to the changes of EBV that occurred in the H line. The regression coefficients of the average EBV on the generations were −8.01, −0.80, −2.01, and 0.00, for BW, BWG, FC, and FCR, respectively, in the H line. Among all the traits, only BW in the H line had a slope (P < 0.05). Similarly, the regression coefficients were −0.506, −0.136, −0.096, and 0.00 for BW, BWG, FC, and FCR, respectively, in the L line. None of the regression coefficients in the L line was different from zero (P > 0.05).
Estimation of Genetic Parameters of the Related Traits Using Pooled Data Sets The genetic correlations (rA) between PBA and BW, BWG, FC, and FCR are presented in Table 4 and the heritabilities (h2) for BW, BWG, FC, and FCR estimated with different pooled data sets are listed in Table 5. The heritabilities of BW, FC, and FCR were consistent in E1 to E3. The negative genetic correlations of PBA with BW, and FC were stronger in E2 than in E3, and the values in E1 were between those of E2 and E3. The negative genetic correlations between FCR and PBA were high in all 3 estimations. The genetic correlation between PBA and BWG was positive [0.24 ± 0.14 (SE)] in E3, however, this correlation was zero and very low in E1 and E2, respectively. The phenotypic data exhibited fluctuations across generations and still had not formed a trend after selecting for 3 generations. This could be attributed to environmental factors, such as variation in feed ingredients, excreta grinding, and temperature. It also means that, in a shortterm selection project when genetic progress is relatively small compared with the environmental effects, we can-
Downloaded from http://ps.oxfordjournals.org/ at Georgian Court University on March 10, 2015
the studied traits, I is an identity matrix, and R is the (co)variance matrix for the corresponding residual effects. Generations and hatch groups within generations were combined into 24 generation-hatch groups. Sex and generation-hatch groups were included in the model as fixed effects. The formation of A−1 was based on Henderson’s (1975) and Quaas’s (1976) methods. The calculation of BLUP and variance components was accomplished using the average information algorithm for restricted maximum likelihood (REML) procedure (Johnson and Thompson, 1995) with an AiREML program (Misztal et al., 2002). Convergence was considered to have been reached when
540
ZHANG ET AL.
not expect to infer from the phenotypic trend. Evaluation of the correlated responses should therefore be determined from both line comparison, and the estimation of breeding values. When a comparison of the breeding values at G3 was made, the differences of the EBV between H line and L line were close to the CRC values for BW, BWG, and FC. They were −23.86 vs. −26.80 g for BW; −2.05 vs. −3.40 g for BWG; and −5.85 vs. −4.98 g for FC. However, at G1 and G2, the line differences of EBV deviated from the CRC values for BWG and FC. This inconsistency could be attributed to experimental errors and genetic drift that were not accounted for by the fixed model for obtaining CRC. This also means the least-square analysis based on line comparisons sometimes did not indicate the true genetic changes that had occurred in a short-term selection project (Sorensen and Kennedy, 1984; Kennedy, 1990; Walsh, 2004). A trend in FCR for either CRC or EBV criteria could not be ascertained. This implies that in the short-term, selection for PBA did not affect feed efficiency. This result was confirmed by the small genetic correlations between PBA and FCR estimated with the records of the base
population. Theoretically, simple linear equations can be established by regressing EBV on generations for predicting the genetic trends of the related traits (Falconer and Mackey, 1996; Walsh, 2004), but in the current study, selection was undertaken only for 3 generations, and the genetic changes were relatively small and not stable for most traits, and consequently only the equation for BW in the H line had a slope coefficient (P < 0.05). Therefore, data from more generations of selection will be needed for establishing useful prediction equations for BWG and FCR. The profile of the genetic correlations of PBA and the correlated traits estimated using the pooled data set did not follow any pattern. In the estimation using the data of the base population exclusively, the genetic correlations of PBA with BW, BWG, and FC were negative and moderate (Zhang et al., 2003). This was in agreement with genetic trends in both the H and L lines. However, when information of the selected generations (G1 to G3) was combined, the relationship tended to become weaker, even reversed in direction, and showed line dimorphism (Table 4).
Downloaded from http://ps.oxfordjournals.org/ at Georgian Court University on March 10, 2015
FIGURE 2. Genetic trends for BW, body weight gain (BWG), feed consumption (FC), and feed conversion ratio (FCR) in chicken populations selected for high (H line) or low (L line) phytate phosphorus bioavailability.
541
CORRELATED RESPONSES TO SELECTION FOR PHYTATE PHOSPHORUS TABLE 5. Genetic variance (σ2A) and heritability (h2) (± SE) of BW at 4 wk, BW gain (BWG), feed consumption (FC), and feed conversion ratio (FCR) estimated using different pooled data sets Estimation 1 Trait BW BWG FC FCR
σ2A 1,080.89 32.35 92.75 0.0
± ± ± ±
Estimation 2 h2
78.62 8.12 13.44 1.00
0.58 0.27 0.41 0.08
± ± ± ±
σ2A 0.03 0.06 0.05 0.02
1,114.03 31.61 85.40 0.02
± ± ± ±
Estimation 3 h2
84.27 7.68 13.68 0.00
0.60 0.26 0.38 0.09
± ± ± ±
σ2A 0.03 0.06 0.05 0.01
1,003.51 30.41 94.02 0.01
± ± ± ±
h2 94.82 10.16 16.21 0.00
0.57 0.26 0.43 0.09
± ± ± ±
0.04 0.08 0.06 0.02
1 Estimation 1 = estimated using the whole pooled data; estimation 2 = data from generation 0 and generations 1 to 3 of the high PBA line; estimation 3 = data from generation 0 and generations 1 to 3 of the low PBA line.
contributing factors for the moderately negative genetic correlations of PBA with BW, BWG, and FC, which were estimated using the data of the base population exclusively (Zhang et al., 2003). Selection for PBA could change the frequency of the allele(s) disproportionately in the divergent lines leading to the asymmetric responses in the correlated traits. It is also possible that genetic homeostasis between PBA and the correlated traits were different in the H and L lines. In summary, from the current investigation it can be inferred that direct selection for PBA can consequently led to concurrent changes in BW, BWG, FC, and FCR in the H line, however, similar changes in the L line was less appreciable.
ACKNOWLEDGMENTS The authors thank Ignacy Misztal and Shogo Tsuruta, Department of Animal and Dairy Science, University of Georgia (Athens, GA) for their permission to modify and use their program, and Cheryl Pearson Gresham, Mohamet Komberi, Donna Slavin, and Elaine Foster, Department of Poultry Science, University of Georgia, for their technical assistance.
REFERENCES Aggrey, S. E., W. Zhang, R. I. Bakalli, G. M. Pesti, and H. M. Edwards, Jr. 2002. Genetics of phytate phosphorus bio-availability in poultry. Pages 277–279 in Proceedings of the 7th World Congress on Genetics Applied to Livestock Production. Montpellier, France. Bovenhuis, H., H. Bralten, M. G. B. Niuwland, and H. K. Parmentier. 2002. Genetic parameters for antibody response of chickens to sheep red blood cells based on a selection experiment. Poult. Sci. 81:309–315. Dodenhoff, J., L. D. Van Vleck, S. D. Kachman, and R. M. Kock. 1998. Parameter estimates for direct, maternal, and grandmaternal genetic effects for birth weight and weaning weight in Hereford cattle. J. Anim. Sci. 76:2521–2547. Dunnington, E. A., and P. B. Siegel. 1996. Long-term divergent selection for eight-week body weight in White Plymouth Rock chickens. Poult. Sci. 75:1168–1179. Edwards, H. M., Jr. 1983. Phosphorus I. Effect of breed and strain on utilization of sub-optimal level of phosphorus in ration. Poult. Sci. 62:77–84. Falconer, D. S., and T. F. C. Mackay. 1996. Introduction to Quantitative Genetics. Longman Group, Essex, UK. Fredeen, H. T., and H. Mikami. 1986. Mass selection in a pig population: Correlated changes in carcass merit. J. Anim. Sci. 62:1654.
Downloaded from http://ps.oxfordjournals.org/ at Georgian Court University on March 10, 2015
In the estimation (E2) using the combined data of G0 and the H line, the genetic correlation between PBA and BW was moderately negative and this was consistent with the genetic trends in the H line (Table 4). In the estimation (E3) using the combined data of G0 and L line, the genetic correlation between PBA and BW was almost negligible and this was consistent with the lack of genetic changes in L line. A similar phenomenon was observed for FC. The genetic correlation between PBA and BWG in the estimations using pooled data sets (E1 to E3) was unexpected and not in agreement with the genetic trend profile. The selection showed a negative genetic trend in the H line, yet the genetic correlation was negligible in the estimation (E2) using the combined data from G0 and the H line (Table 4). The large negative genetic correlation between PBA and FCR in E1 to E3 is difficult to interpret. It seems that the genetic covariance of PBA and FCR was unstable because their genetic variances were small and their measurement was subject to substantial errors in the short-term selection experiment. The mechanism underlying the inconsistent estimates of genetic correlations using different data sets is still not clear. However, Dunnington and Siegel (1996) suggested that changes in resource allocation, or prioritization of resource use of demands placed on artificially selected chickens may be responsible for changes in the genetic covariance among traits. Resource allocation or prioritization of resource use could be the underlying factors behind the asymmetric correlated responses in BW, BWG, and FC. Another possible explanation for the asymmetric correlated responses could be the involvement of major gene(s). Some major genes have modest effects on the associated metric traits, and do not significantly influence the distribution profiles of the phenotypic and genetic values when the frequencies are low. For example, a recessive and sexlinked dwarfing allele, dwB, in poultry was found to be associated with only a 10% BW reduction, smaller than a standard deviation (Shiva-Prasad and Jaap, 1977). Even though the inheritance of PBA did not deviate from an additive model of many loci (Zhang et al., 2003, 2005), it is still possible that a recessive allele(s) with a “large effect” and a low frequency in the base population in addition to genes of small effect controlled PBA. The pleiotropic effects of the allele(s) may be associated with high PBA and low growth rate. This could be one of the
542
ZHANG ET AL. Ravindran,V., W. L. Bryden, and E.T. Kornegay. 1995. Phytates: Occurrence, bioavailability, and implications in poultry nutrition. Poult. Avian Biol. Rev. 6:125–143. SAS Institute. 1998. SAS version 8.12. SAS Institute Inc., Cary, NC. Sewalem, A., K. Lillpers, and M. Wilhemson. 1998. Genetic trends of production and reproduction traits in White Leghorn lines selected for production traits. Pages 286–289 in Proceedings of the 6th World Congress on Genetics Applied to Livestock Production, Armidale, Australia. Schulman, N., M. Tuiskula-Haavista, L. Siitonen, and E. A. Ma¨ntysaari. 1994. Genetic variation of residual feed consumption in a selected Finnish egg-layer population. Poult. Sci. 73:1479–1484. Shiva-Prasad, H. L., and R. G. Jaap. 1977. Egg and yolk production as influenced by liver weight, liver lipid and plasma lipid in three strains of small bodied chickens. Poult. Sci. 73:1479–1484. Sorensen D. A., and B. W. Kennedy. 1984. Estimation of genetic variances from unselected and selected populations. J. Anim. Sci. 59:1213–1223. Stuard, A., and J. K. Ord. 1994. Kendall’s Advanced Theory of Statistics. Vol. 1. Distribution Theory. 6th ed. Halsted Press, New York, NY. Tufvessen, M., B. Tufvessen, T. von Schantz, K. Johansson, and M. Wilhelmson. 1999. Selection for sexual male characters and their effects on other fitness related traits in White Leghorn chickens. J. Anim. Breed. Genet. 116:127–138. Walsh, B. 2004. Subject: Evolution and Selection of Quantitative Traits. Pages 165–237. http://nitro.biosci.arizona.edu/ zbook/volume_2/vol2.html. Accessed July 1, 2004. Zhang, W., S. E. Aggrey, G. M. Pesti, H. M. Edwards, Jr., and R. I. Bakalli. 2003. Genetics of phytate phosphorus bioavailability: Heritability and genetic correlations with growth and feed utilization traits in a randombred chicken population. Poult. Sci. 82:1975–1979. Zhang, W., S. E. Aggrey, G. M. Pesti, R. I. Bakalli, and H. M. Edwards, Jr. 2005. Genetic analysis on the direct response to divergent selection for phytate phosphorus bioavailability in a randombred chicken population. Poult. Sci. 84:370–375.
Downloaded from http://ps.oxfordjournals.org/ at Georgian Court University on March 10, 2015
Garcia, M. L., and A. Baselga. 2002. Estimation of correlated response on growth traits to selection in litter size of rabbits using a cryopreserved control population and genetic trends. Livest. Prod. Sci. 78:91–98. Henderson, C. R. 1973. Sire evaluation and genetic trends. Pages 10–41 in Proceedings of Animal Breeding and Genetics Symposium in honor of Dr. J. L. Lush. American Society of Animal Science, Champaign, IL. Henderson , C. R. 1975. Rapid method for computing the inverse of a relationship matrix. J. Dairy Sci. 58:1727–1730. Henderson, C. R. 1984. Applications of Linear Models in Animal Breeding. University of Guelph, Ontario, Canada. Heuser, G. F., L. C. Norris, J. McGinnis, and M. L. Scott. 1943. Further evidence of the need for supplementing soybean meal chick rations with phosphorus. Poult. Sci. 22:269–270. Johnson, D. H., and R. Thompson. 1995. Restricted maximum likelihood estimation of variance components for univariate animal models using sparse matrix techniques and average information. J. Dairy Sci. 78:423–447. Kaplon, M. J., M. F. Rothschild, P. J. Berger, and M. Healey. 1991. Genetic and phenotypic trends in Polish Large White nucleus herds. J. Anim. Sci. 69:551–558. Katle, J., and N. Kolstad. 1991. Selection for efficiency of food utilization in laying hens: Direct response in residual food consumption and correlated responses in weight gain, egg production and body weight. Br. Poult. Sci. 32:939–953. Kennedy, B. W. 1990. Mixed model methodology in analysis of designed experiments. Pages 76–94 in Advances in Statistical Methods for Genetic Improvement of Livestock. D. Gianola, and K. Hammond, ed. Springer-Verlag, Berlin, Germany. Kuhlers, D. L., and S. B. Jungst. 1993. Correlated responses in reproductive and carcass traits to selection for 200-day weight in Landrace pigs. J. Anim. Sci. 71:595–601. Misztal, I., S. Tsuruta, T. Strabel, B. Auvray, T. Druet, and D. H. Lee. 2002. BLUPF90 and related programs (BGF90). Pages 743–744 in Proceedings of the 7th World Congress on Genetics Applied to Livestock Production, Montpellier, France. Mrode, R. A. 1996. Linear Models for the Prediction of Animal Breeding Values. CABI, Wallingford, UK. Quaas R. L. 1976. Computing the diagonal elements and the inverse of a large numerator relationship matrix. Biometrica 32:949–953.