Direct and correlated responses to selection in two flocks of White Leghorn Summary

Data on 17,588 pulletes of two strains selected for egg number and egg weight alongwith a control line were examined over seven generations. Significant desirable realized genetic gains/generation were ooserved for egg number to 40 weeks (2.18 and 2.23 eggs) and egg weight (0.81 and 0.45g) in both the strains. Significant correlated responses were also observed in age at first egg (-1.54 and -1.17 d); 20 and 40 week body weights (8.8 to 19.0). Significant desirable responses were also observed for egg mass (175 and 151g), rate of lay (1.0 and 1.37%), egg produetion efficiency (0.08 and 0.07) and efficiency index (1.09 and 1.03 g/d/kg) Natural selection did not seem to play any role. Inbreeding of 0.28 to 0.45% per generation did not appear to be of any significance in affecting the performance and heritabilities. The additive genetic and phenotypic variances and heritabilities remained stable over generations. Realized heritabilities varied between 0 18 and 0.29 for egg number and between 0.44 and 0.66 for egg weight and indicated the effectiveness ofthe criterion of selection.


Introduction
The economic gain in layer chickens depends on a number of mutually associated traits some of which are antagonistically correlated (FAIRFULL and GOWE, 1990).A number of selection experiments for increasing part-year egg produetion have yielded positive responses in egg number but also invariably produced concomitant reduction in egg weight (recent ones being those ofAYYAGARI et al., 1983;JORJANI et al., 1993;SHARMA et al., 1998).There are few reports of selection for the simultaneous improvement of egg produetion, egg weight and other traits in layer Stocks undergoing long-term selection with sufficiently large population sizes and/or a suitable control population (GOWE and FAIRFULL, 1985;JORJANI et al., 1993).A selection experiment for the simultaneous improvement of egg produetion, egg weight and some other traits in two White Leghorn populations has been in progress at this Institute.Results of selection upto 12* and 9 th selected generations had been reported eariier (BRAH et al., 1993) but the absence of a control population did not permit the Separation of environmental and genetic trends.The present paper reports the results for the selected and unselected traits for the subsequent seven generations in which a control population was contemporarily reared.

Materials and Methods
Populations: Two single Comb White Leghorn populations (PL 1 and PL 2) undergoing selection and a pedigreed control line (PL 3) maintained at the University farm provided the experimental material.Strain PL 1 was originally derived from a reputed commercial stock imported from USA.Strain PL 2 was derived from the wellknown pure-bred 'Mount Hope' stock of USA.The control line (PL3) was constituted from strain PL2, after it had undergone 8 generations of selection, by randomly choosing cockerels and pullets.The average number of progeny (pullets) tested, the effective number of sires and dams and the effective population size during the course of the present study are given in Table 1.The population structure in the eariier generations was also similar (BRAH et al., 1993).Main emphasis in selection was laid on egg number upto 40 weeks of age using a family index that took into account the individual performance plus dam and sire family averages for pullets, and dam and sire family averages for cockerels (OSBORNE, 1957a, b).Selection for egg weight (independent culling levels) had also been practiced from the ineeption of the selected strains.The strains had also undergone mild selection for liveability, fertility, hatchability and adult body weight in previous generations.For reproducing the control line (PL 3) as far as possible each sire contributed a sire and each dam contributed a dam as breeders to minimize genetic drift and inbreeding.Füll-and half-sib matings were avoided in all the lines to keep inbreeding low.The chicks were obtained in 2-3 hatches every year in the months of March and April and were reared under Standard nutritional and managemental conditions.

Traits:
The basic traits recorded on each pullet were: age at first egg (AFEV egg number to 40 weeks of age (EN40); body weights at 20 and 40 weeks of age (BW20 and BW40) and egg weight (EW) based on average of 6-8 eggs measured between 36-38 weeks of age.Derived traits included rate of lay (ROL); egg mass to 40 weeks of age (EM); ratio of 40 to 20 week body weight (Ratio BW); egg weight as % of 40week body weight (EW % BW); egg mass per functional day (EMFD); egg produetion etiiciency i.e. egg mass in grams divided by body weight at 40 weeks in grams (EPE) and efficiency index i.e. egg mass per fuctional day in grams divided by body weight at 40 weeks in kilograms (EINDEX).

Statistical Analysis:
Hatch adjusted Least-square means (HARVEY, 1988) were used for estimation of phenotypic and genetic changes.The realized genetic gain per generation was estimated as deviation ofthe selected line means from the control line means regressed over generations.The effective number ofsires and dams were used tor the computation of effective population size and inbreeding as per GOWE et al (1959).The effective selection differentials were calculated by weighting each parent by the number of offspring available at 40 weeks of age.The standardised selection differential for the selected traits was obtained by dividing selection differential with the phenotypic Standard deviation.Realized heritability was obtained as (i) simple regression of response on cumulative selection differential (b RS ) and (ii) the ratio of cumulative response (CR) to cumulative selection differential (CS) for the selected traits (HILL, 1972).The heritabilities were calculated from full-sib analysis as per the procedure outlined by BECKER (1984).The mean heritability across generations was obtained as simple average ofthe seven estimates.To calculate the Standard error (SE) of h the estimate of each year was treated as a single Observation and SE was calculated from the variability among these estimates (GOWE and FAIRFULL, 1985).

Results and Discussion
Control Population: The environmental trends in the control line, for different traits are presented in Table 3, and Figures 1 and 2. Regression of means on generations were negative and significant for some traits but non-significant for others.Significant systematic changes in the performance ofthe control for a number of traits had been reported in several selection experiments (KOLSTAD, 1980;JORJANI et al., 1993).The Performance of the control showed fluetuations over generations (Fig. 1 and 2).However, the pattern of changes in means of control over generations fairly matched with those observed for the two selected lines indicating possibly these to be of the environmental origin.The selection differentials computed due to random selection of control hne breeders were essentially zero thus further strengthening the lack of a genetic trend in the control line.Rate of inbreeding was very low (0.28%) and aecrued to be 1.96% over seven generations and is therefore unlikely to have affected the Performance and variance to any significant degree.The estimates of additive genetic variance and heritabilities did not change over generations (Table 4).Keeping in view  fi^notvntrTJ r f0 T nC ti°f SeleC ', ed Snd COntr °' ]i " eS (UPPer) and 8enetiC Chan 8 e < l0Wer ) ta selected lines (Pänotypische Leistungsentwicklung m den Selektions-und Kontrollpopulationen) the size of the breeding population maintained and the mating procedure adopted to mmimize the genetic drift and inbreeding, coupled with the lack of changes in genetic variance and heritabilities, there appeared no evidence of a genetic change in the control population.

Mean performance and genetic gains:
The mean phenotypic performance of the selected and control lines and the means of the selected lines expressed as deviations from the control means for some important traits are presented in Fig. 1 and 2 while the estimates of genetic gains for all the traits are given in Table 3.The means for egg number to 40 weeks of age for the three lines are given in Table 2. Selected traits: The realized genetic gain for egg number to 40 week of age was significantly positive in both the strains (2.18 and 2.23 eggs/generation).The significant improvement observed for egg number in the selected lines resulted due to desirable correlated responses in age at sexual maturity and rate of lay.This was because of the fact that egg number selection is also negative selection for age at first egg and positive selection for rate of lay.Reduction in sexual maturity and improvement in laying intensity due to selection for egg number alone has been observed in a number of experiments (FAIRFULL and GOWE, 1990).In the few selection experiments that included simultaneous selection for egg number and egg weight (KOLSTAD, 1980;LILJEDAHL and WEYDE, 1980;SORENSEN et al., 1980), reduction in age at first egg to the extent of 2 days/generation was observed while the rate of lay showed either an increase or remained static.As a consequence of mild selection applied for egg weight, it exhibited significant genetic improvement in both the strains (0.63 and 0.81 g/generation).Eariier results of selection experiements for a combination of egg number and egg weight have also reported significant increase of 0.2 to 0.4 g/generation over short-term periods (FAIRFULL and GOWE, 1990.Correlated traits: Significant positive correlated responses were observed for 20-and 40-week body weights in both the strains.A trend for increase in housing and mature body weight with selection for egg produetion plus egg weight probably due to the influence of egg size has been observed in most of the lines in such selection experiments (KOLSTAD, 1980;LILJEDAHL and WEYDE, 1980;SORENSEN et al., 1980).This is in contrast to the results of single-trait selection for egg produetion (AYYAGARI et al., 1983;SHARMA et al., 1998) in which housing and mature body weights showed a downward trend.The ratio of 40-to 20-week body weight did not change indicating that both the body weights were equally altered.Egg weight expressed as % of body weight did not change over generations which implied that the increase in mature egg weight was associated with a corresponding improvement in adult body weight.Egg mass output showed significant genetic improvement in both the lines (Fig. 2 and  Table 3).The rate of genetic improvement was more in PL 1 because of its higher rate of genetic improvement for egg weight as compared to PL 2. The magnitude of responses for egg mass in the present study match fairiy well with those reported for short-term selection experiments for egg produetion plus egg weight (KOLSTAD, 1980;LILJEDAHL and WEYDE, 1980) but are slightly higher than those reported by JORJANI et al. (1993) over 10 generations of selection.Egg mass per functional day also showed significant genetic improvement (Table 3).The direct measurement of efficiency of feed conversion in layers is cumbersome and expensive.NORDSKOG et al. (1974) have shown that egg produetion efficiency (egg mass/body weight) and efficiency index (egg mass/functional day /kg body weight) are useful indirect measures of feed efficiency for egg produetion.Efficiency index has also been found to be correlated with the margin of profit over feed cost in commercial Stocks (SHALEV, 1980).Both these measures of efficiency showed significant genetic ATVATT 11 * °f essentiall y similar magnitude in both the selected strains.AY YAGARI et al. (1983) in a single-trait egg produetion experiment failed to observe any significant improvement in efficiency index.

Selection differentials:
The effective selection differentials for egg number varied between 8.1 and 11.7 eggs in PL 1 and between 9.3 and 12.2 eggs in PL 2 over generations.These values match fairiy well with those reported by AYYAGARY et al ^oSl and J S ?ARMA Ct aL °"8) fr0m sin 8 le -*ait selection, and also with those of GOWE and FAIRFULL (1985) from multi-trait selection experiment.The effective selection differentials for egg weight varied between 0.87 and 1.64 g in PL 1 and between 0 16 and 1.42 g in PL 2 over generations.The regression coefficients of selection differential on generation number were negative for egg number but !~. R 2 onl y ^ P°s itive a ^d significant for egg weight in both the strains Uable 4).This implied a tendency for selection for egg number to decrease gradually with increased emphasis being laid on the improvement of egg weight over generations.The selection emphasis for egg number was marginally higher in PL 1 as compared to PL 2. However, egg weight received about 47% more selection weightage in PL 1 as compared to PL 2. A high degree of agreement between the expected and effective selection differentials suggested that natural selection did not influence the artificial selection for egg number and egg weight (Table 4).The direction of the selection differentials of correlated traits was in accordance with the direction of their realized genetic gains (Table 3).Inbreeding: Inbreeding averaged to be 0.45; 0.44 and 0.29% per generation for the selected strains PL 1 and PL 2 and the control line (PL 3) respectively for the seven generations reported herein (Table 1).Identical rate of inbreeding prevailed during the first phase of selection in these strains (BRAH et al., 1993).The level of inbreeding upto the end ofthe present study (i.e.19* and 16* generations in PL 1 and PL 2) thus accrued to be 8.25 and 7.54%.These levels of inbreeding particularly with the avoidance of matings of relatives are unlikely to have influenced the performance of the selected lines.Eariier reports suggest that 10-15% or even higher levels of inbreeding did not much affect the performance ofthe selected lines (NORDSKOG et al., 1974;ABLANALP, 1990).The drift and error variance per generation, computed as per HILL (1980), were low suggesting that these factors were well controlled.Variance and heritability estimates: The two principal traits under selection showed fluctuations over generations in heritabilities and variances (Fig. 3) but there seemed no time-trend over generations with the regression coefficients being non-significant (Table 5).The heritabilities and variances for the correlated traits also remained unchanged (Table 5).The mean heritabilities ofthe control line for the selected and the correlated traits were similar to those of the two selected strains.The heritabilities estimated in the eariier generations (1-12 and 1-9) for the principal traits under selection as well as for other traits (BRAH et al., 1993) were of similar magnitude as observed for these seven generations.Theoretically selection is expected to reduce genetic Variation and heritability though the magnitude of its effect will be dependent on the intensity of selection and the level of heritability (FALCONER, 1989).The selection intensities in the present study were moderate for egg number (0.56 and 0.59 o-p) and low for egg weight (0.39 and 0.29 o>) in the two strains.The lack of reduction in heritability and genetic Variation may thus in part be attributable to moderate intensity of selection applied and also due in part to mutations which have been shown to play a significant role in long-term selection experiments (FRANKHAM, 1980).

Realized heritabilities vis a vis effectiveness of selection:
Selection in the present study for egg number was based on multi-sources of information and selection was truly not of the truncation-type since independent culling levels were also used for the improvement of egg weight.Therefore, the realized heritabilites reported in Table 6 may not be considered valid estimates of heritability of individual values for egg number in these strains.However, the realized heritabilities provide the most useful empirical description of the effectiveness of selection (FALCONER, 1989).The values of realized heritabilities averaged over the two methods and strains worked out to be 0.23 and 0.54 for egg number and egg weight respectively.The high degree of correspondence between the realized and estimated heritabilities for egg weight implied that the magnitude of the realized genetic gains was as per the expectation based on theory.The excess of realized heritability of egg number over the sire-component heritability by 28% implied that use of familiy averages in addition to individual's own performance, improved the rate of genetic progress over individual selection.With an average effective selection differential/generation of about 10.1 eggs and an average realized heritability of 0.23, the expected genetic gain of 2.32 eggs is close to the realized gain of 2.20 eggs (Table 3).For egg weight, with average selection differential of 1.07 and average realized heritability of 0.54 across the two strains, the expected gain of 0.58 g also matched with the average realized gain of 0.63 g for the two strains.

Fig. 1 :
Fig. 1: Mean phenotype performance of selected and control lines (upper) and genetic change (lower) in selected lines (Pänotypische Leistungsentwicklung in den Selektions-und Kontrollpopulationen)

Table 3
Regression of response on generation numbers (b±SE) (Direkte und indirekte Selektionserfolge je Generation)

Table 6
Estimates of realized h* for egg number and egg weight (Realisierte Heritabilitäten für Eizahl und -gewicht)