Direct and Correlated Responses to Multitrait, Divergent Selection for Immunocompetence1

Direct and Correlated Responses to Multitrait, Divergent Selection for Immunocompetence1 R. P. KEAN Department of Animal Science, Iowa State University, Ames, Iowa 50011 A. CAHANER Department of Genetics, Faculty of Agriculture, The Hebrew University of Jerusalem, P.O.B. 12, Rehovot 76100, Israel A. E. FREEMAN and S. J. LAMONT2

ABSTRACT Leghorn lines had been selected for an immunocompetence index based on four traits: antibody production to Mycoplasma gallisepticum (MG) and Pasteurella multocida (PM) vaccines, reticuloendothelial clearance of colloidal carbon (CCA), and cell-mediated, wing web response to phytohemagglutinin (PHA). The purpose of this study was to produce replicated lines of chickens with divergent levels of multitrait immunocompetence by index selection. The objectives of analyses of Generations 5 to 7 of this study was to characterize these lines with respect to immune-response traits, correlations among these traits, and correlated responses in other important production traits. Differences (P < .05) existed between the lines selected for high or low immune response and between the two replicates in mean breeding values and in individual immune-response traits. Averages of heritability estimates, weighted by number of offspring and pooled across three generations (two cycles of selection), estimated by using sire variance components and parentoffspring correlations were, respectively, .16 and .09 for the index, .31 and .08 for MG, .21 and -.02 for PM, .06 and .05 for CCA, and .08 and .12 for PHA. Realized heritabilities (response divided by effective selection differential) pooled across the two selection cycles, were .19 and .11 for the index, .06 and -.01 for MG, .44 and .32 for PM, 1.52 and -1.21 for CCA, and .48 and .15 for PHA, for Replicates 1 and 2, respectively. Phenotypic correlations among traits were generally small, and several estimates were negative. Estimates of genetic correlation varied widely. Juvenile and adult body weights, age of first egg, 32-wk egg weight, and rate of egg production were analyzed to evaluate effects of selection on these traits of direct economic importance. Very few differences were noted. (Key words: index, selection, immune response, White Leghorn, chicken) 1994 Poultry Science 73:18-32

INTRODUCTION Received for publication June 14, 1993. Improving disease resistance without Accepted for publication September 23,1993. actually challenging animals with a disijournal Paper Number J-15342 of the Iowa e a e Ja e n.t {-desirable Because an 1S Agriculture and Home Economics Experiment Station, ? S ^esiraDie. Because an Ames, IA 50011. Projects Numbers 2237 and 2351. animal s response to the presence of a 2To whom correspondence should be addressed, disease agent usually involves some form 18

Downloaded from http://ps.oxfordjournals.org/ at Kainan University on April 9, 2015

Department of Animal Science, Iowa State University, Ames, Iowa 50011

EFFECT OF SELECTION FOR MULTTTRAIT IMMUNOCOMPETENCE

tion between resistance to viral and protozoal diseases in lines selected for antibody production to SRBC but found negative correlation between anti-SRBC antibody production and resistance to bacterial diseases. Mallard et al. (1992) characterized the immune-responsiveness of a random-bred population of Yorkshire pigs by using 14 indicator traits. Five traits [antibody production against hen egg white lysozyme (HEWL), total serum IgG, delayedtype hypersensitivity to a purified protein derivative of Bacillus (PPD), blastogenic response to concanavalin A (ConA), and monocyte killing of Salmonella typhimurium] were then chosen for multitrait selection. After one generation of selection based on a combined estimated breeding value (EBV), high and low lines differed significantly with respect to HEWL, PPD, and ConA, and also with respect to total EBV. Heritability estimates ranged from 0 to .25. The ConA and HEWL were positively correlated in the base population and no other significant correlations among these five traits were reported. Experiments with selection for immune response have yielded varied results with respect to correlations with other important economic traits. In pigs, Meeker et al. (1987) found a negative correlation between growth traits and immune responses to both Bordetella bronchiseptica and pseudorabies virus vaccines but found no correlation between the two antigens and backfat thickness. Researchers working with dairy cattle (DetiUeux et al, 1991; Kehrli et al, 1991; Weigel et al, 1992) have used a somewhat reversed approach by selecting lines divergent in milk production and then comparing the lines' immune responsiveness. Several negative correlations with milk production have been found in these studies. In chickens, Siegel and Gross (1980) and Siegel et al. (1982) reported a negative correlation between body weight and antibody production in lines divergently selected for antibody production to SRBC. Van der Zijpp (1983) also reported negative correlation between antibody production to SRBC and body weight, and no other significant correlations with produc-

Downloaded from http://ps.oxfordjournals.org/ at Kainan University on April 9, 2015

of immune response, selection for immunocompetence has been suggested as an indirect method of improving disease resistance (Gavora and Spencer, 1983; van der Zijpp, 1983; Warner et al, 1987). Selection experiments using a variety of agents to elicit an immune response have been conducted (Bacon, 1992). Biozzi et al. (1979) reported selection experiments to several antigens in mice. The results of these experiments suggested that, in some instances, selection for increased antibody production to one antigen (for example SRBC) improved the responses to several other antigens. Antibody production against SRBC has also been used in selection experiments in chickens (Siegel and Gross, 1980; van der Zijpp et al, 1983; Martin et al, 1990; Pinard et al, 1992). Pevzner et al (1981a) reported on selection for antibody response to the synthetic amino acid polymer, GAT and also for antibodies against Salmonella pullorutn (Pevzner et al, 1981b). Antibody production against Newcastle disease virus and Escherichia coli (Pitcovski et al, 1987) and against E. coli alone (Leitner et al, 1992) have also been used in selection experiments. Karakoz et al. (1974) selected for cell-mediated immune response by using purified protein derivative of tuberculin. Although these and other experiments have been conducted by selecting for immune response to some specific agent (or specific facet of immune response), selection for general immunocompetency, combining multiple facets of immune response, might be more desirable (Gavora and Spencer, 1978; van der Zijpp, 1983). Successful selection for different components of immune response simultaneously requires a knowledge of each individual trait and the interactions between the different facets of immune response and associations with disease resistance and other important biological traits. Biozzi et al. (1979) reported a negative correlation between antibody production and phagocytic ability in lines of mice divergently selected for SRBC response. Most of the other correlations found by Biozzi et al. (1979) between different measures of different facets of immune response were relatively small. Gross et al (1980) found positive associa-

19

20

KEAN ET AL.

Downloaded from http://ps.oxfordjournals.org/ at Kainan University on April 9, 2015

tion traits were found. Although an- tion, chickens with the highest ranking tagonistic correlations with important eco- breeding values within each sire family nomic traits could diminish selection within each high-immune-response line pressure and, therefore, slow improve- and chickens with the lowest ranking ments in immune response traits, simul- breeding values within each sire family taneous selection for negatively correlated within each low-immune-response line were chosen as breeders for the next traits is still possible (Falconer, 1989). The overall purpose of this selection generation. Ten sires and 40 dams were project was to develop replicated lines of used as breeders in each of the four lines in chickens divergent in levels of multitrait each generation. Immunocompetence immunocompetence and to characterize parameters of the base generation and first the genetic parameters of the traits in- two selected generations have been volved. Cheng et al. (1991) reported reported previously (Cheng et al, 1991). All chickens in each generation were parameters from the base population and raised in floor pens until 20 wk of age and in the first two selected generations of chickens in this project. This report presents individual cages after 20 wk of age. They results from the fifth, sixth, and seventh were provided ad libitum access to feed generations of these lines. Genetic meeting all NRC (1984) requirements and parameters of the selection index and the water at all times. four individual traits that are used in that index, as well as associations between the Immunocompetence Measures immune response traits and production The immunocompetence measures used traits, are presented. in this project have been described in detail (Cheng et al, 1991). Brief descriptions MATERIALS AND METHODS follow. Humoral immune response was measured as antibody production to two vaccines. Vaccines against Mycoplasma galChickens lisepticum (MG bacterin) 3 (MG) and Divergent selection for multitrait im- Pasteurella multocida (FC bacterin)3 (PM) munocompetence has been conducted in a were injected at 6 wk of age. Serum group of White Leghorn chickens derived samples, collected at 9 wk of age, were from the Ottawa Control Strain 7 (Gowe analyzed for specific antibodies by using and Fairfull, 1980). High-responding half- commercial ELISA4 kits according to the siblings from 10 sires and 36 dams were manufacturer's directions. Antibody used to produce a high-immune-response production was calculated as a corrected line (1H) and corresponding low- ratio of the individual's sample and a responding half-siblings from the same 10 positive control (S - N)/(P - N), where S = sires and 36 dams were used to produce a sample; N = negative control; and P = low-immune-response line (1L) in Replicate positive control; per kit instructions. The 1. Similarly, half-sib offspring from another positive and negative controls were chicken 10 sires and 37 dams were used to produce a sera included in the commercial kits and high-immune-response line (2H) and a low- were tested in duplicate on each plate of immune-response line (2L) in Replicate 2. samples. The MG and PM responses are Matings for future generations were within expressed as a ratio of sample to the lines, based upon breeding values (esti- positive serum. Sera collected before initial mated by using the index described below), injection were also tested to evaluate any possible previous exposure and response to and sibling matings were avoided. All progeny (approximately 500) were these antigens. tested in each generation. In each generaCell-mediated immune response was measured by using wing web response to phytohemagglutinin (PHA-P)s (van der Zijpp, 1983; Lamont and Smyth, 1984) when 3Solvay Animal Health, Charles City, IA 50616. the chickens were 10 to 12 wk of age. Wing 4 Idexx Corp., Portland, ME 04104. 5 web response (PHA), in millimeters, was Fisher Scientific, Itasca, IL 60143.

EFFECT OF SELECTION FOR MULTTTRAIT IMMUNOCOMPETENCE

21

four traits. With no estimates of relative economic worth for each trait, balanced PHA = (postlNJ - prelNJ) - gains in all three facets of immune response (postPBS - prePBS), [1] were deemed desirable. Cheng (1990) used parameter estimates from the base populawhere postlNJ = thickness of test wing 24 h tion and compared different weights for the after injection of PHA; prelNJ = thickness of four traits. He determined that the inverses test wing before injection; postPBS = thick- of the traits' estimated heritabilities equalness of control wing 24 h after injection of ized gains better than did the inverses of the physiological saline (PBS); and prePBS = traits' variances or setting each trait's thickness of control wing before injection weight to unity. Also, for selection in the sixth and seventh generations, for breeders (van der Zijpp, 1983). Reticuloendothelial clearance, a measure of the seventh and eighth generations, of phagocytic ability, was measured by respectively, the weights for each measure using intravenous injection of colloidal of antibody production (MG and PM) were carbon (Glick et al, 1964; Lamont, 1986) as divided in half in an effort to increase modified by Cheng et al. (1991). The CCA selection emphasis for cellular immune response and reticuloendothelial clearance was calculated as in Equation 2 relative to humoral immune response. In „„. _ ln(5 min O.D.) - ln(l min O.D.) addition, all genetic covariances in [G] were set to zero because no previous estimates of covariances for these traits in chickens were [2] available in the literature, and estimates for where 5 min O.D. was the optical density of similar traits in other species were relathe sample drawn 5 min after injection of tively small (Biozzi et al., 1979). Another carbon; 1 min O.D. was the optical density important feature in this selection experiof the sample drawn 1 min after injection, ment was that the parameters used to and the denominator reflected the time calculate index weights were adjusted in between collection of samples (Lamont, each generation to reflect added informa1986; Cheng et al, 1991). The CCA values tion. Heritabilities, genetic variances, and are reported as In of absorbance values (at phenotypic variances and covariances were 675 ran) relative to the individual's normal estimated in each generation by using records from all previous generations and serum. the present generation combined. Methods used to estimate these parameters are Index described below. calculated as

Production Records

Body Weights. Body weights were recorded on all individuals at 2,6,12, and 20 [P] [b] = [G] [a] [3] wk of age and on selected individuals (those chosen to be future breeders) at 32 for [b], a vector of unknown index weights, and 51 wk of age in Generations 5 and 6. where [P] was the phenotypic variance- Body weights at 32 and 51 wk of age were covariance matrix of the four traits (MG, not yet available for Generation 7 at the PM, CCA, and PHA), [G] was the genotypic time of this study's analyses. Egg Production. Egg production variance-covariance matrix of the four traits, and [a] was a vector of "economic" records were collected for hens in Generaweights (Hazel, 1943). Some modifications tions 5 and 6 selected as breeders. Age at were made to the index. Because no esti- first egg (AFE), average egg weight at 32 wk mates of economic values of these traits of age (EGGWT), and egg production in were available, [a] was calculated as the periods 1, 2, 3, 4, and Total (PI, P2, P3, P4, inverse of the heritability of each trait in an and PT, respectively) were recorded. Each effort to balance selection pressure on all production period was 8 wk, and eggs were

Downloaded from http://ps.oxfordjournals.org/ at Kainan University on April 9, 2015

A selection index based on the four described traits was used to rank the chickens in each generation. Index weights were calculated by solving

22

KEAN ET AL.

recorded for 4 d / w k , for a total of 32 d per period. Production was measured on a henhoused basis from 20 wk of age. Egg production records were analyzed by number of eggs produced within each period and then converted to percentages for presentation. Mortality. Mortality was recorded on all chickens to 20 wk of age. The number of accidental deaths was subtracted before analysis. After 20 wk of age, only those chickens selected as breeders 40 males and 160 females) were kept and mortality is, therefore, recorded only on these chickens.

The General Linear Models (GLM) Procedure of SAS® (SAS Institute, 1985) was used to estimate least squares means of breeding values and the four individual traits of immune response in each generation. The model (Model [4]) used for these analyses was Yi,ki = H + Ri + D, + R x D« + sijk + ejjH, [4] where Y is the phenotypic trait (Index, MG, PM, CCA, PHA) of an individual; ft, is the population mean; R is the fixed effect of the ith replicate (1,2), D is the fixed effect of the j * selection direction (H, L), s is the random effect of the k * sire within the i j * replicate and direction, and e is a random residual term. Differences between replicates and between selection directions, and the interaction, were tested for significance by using the mean square for sire as the error term. Variance and covariance components were computed by using the restricted maximum likelihood (REML) method of the VARCOMP Procedure of SAS® (SAS Institute, 1985). Phenotypic variance and covariance components and sire variance components were used to calculate the index weights (by using Equation [3]) and were also used to estimate heritabilities in these lines. Heritabilities were estimated by using three methods. "Narrow sense" heritability (ft2) was estimated by:

8

4 (

<

+

^} %

[5]

[6]

Realized heritabilities were also estimated for each replicate in the two selection cycles analyzed here. Because no control lines were measured, it was necessary to assume that response was symmetric in the high and low selected lines within each replicate (Hill, 1972). Response to selection (AR) was calculated as AR = (H k - Lk) - (Hk_x - Lk_i) [7] where H = high line mean (within a replicate); L = low line mean (within a replicate); and k = generation. Selection differentials (AS) were calculated as AS = (sH k - si*) - (Hk_x - Lk_0 [8] where sH = mean (weighted by number of progeny) of effective breeders in the high line (within a replicate); sL = mean (weighted by number of progeny) of effective breeders in the low line (within a replicate); H = high line mean (within a replicate); L = low line mean (within a replicate); and k = generation. Selection differentials were calculated separately for sires and dams, and the mean of these differentials was used for the final analysis. Realized heritability (ft2) was calculated as AR/AS for each replicate in each selection cycle. The index weights from Generation 6 were used for chickens in all three generations to calculate realized heritabilities.

Downloaded from http://ps.oxfordjournals.org/ at Kainan University on April 9, 2015

fip = *OP-

Statistical Analyses

n2 =

where o 2 = sire variance component and a2. = error variance component (Falconer, 1989). The same model, Model 4, was used for this analysis. Heritabilities were also estimated by using correlations between offspring values and their midparent values. The NESTED Procedure of SAS® (SAS Institute, 1985) was used to estimate within-line correlations. Estimates of correlations (top) w e r e weighted by number of offspring for each parent. Heritability (n p ) was calculated as

EFFECT OF SELECTION FOR MULTITRAIT I M M U N O C O M P E T E N T

Phenotypic and genetic correlations between the index and the four traits were calculated from variance and covariance components. The variance components were estimated by using the REML method of the VARCOMP Procedure of SAS® (SAS Institute, 1985). Again, Model [4] was used. Covariance components (a^) of two traits, x and y, for example, were calculated by using Equation 9

23

cates, between high- and low-responseselected groups, and interactions. Incidence of mortality within replicates and selection directions are reported (in percentages). Differences between replicates, between high- and low-selected groups, and the interaction were tested for significance by using 2 x 2 contingency tables (Snedecor and Cochran, 1989). RESULTS

i-jg

Least squares means for pools of lines [9] by selection direction, Replicates 1 and 2, and their interactions are reported in where or. „ is the variance of the individual Table 1. Comparing high lines and low Ai-y lines, significant differences (P < .05) were sums of the two traits, and a? and <£ are the present with respect to the overall index in variances of x and y, respectively. Genetic Generation 7. Responses to MG differed covariances were calculated by using sire by selection direction in Generations 5 and components of variance, multiplied by 4, in 7 but not in Generation 6. Differences Equation [9]. Phenotypic correlations (r^) were also seen between high-response lines and low-response lines in PM in were calculated by Generation 7 and in PHA in Generation 5. Significant differences were present between replicates in antibody response to f xy = = «•ff-fa. - a [10] MG in Generations 5, 6, and 7, the overall x y index in Generations 5 and 6, in PHA in where
xy =

- * — 2

Downloaded from http://ps.oxfordjournals.org/ at Kainan University on April 9, 2015

24

KEAN ET Ah.

MG PM a = CCA PHA

1.39 2.17 16.13 10.64

TABLE 1. Least squares means of four immune response traits and an index of those four traits, by selection direction and replicate, and levels of significance of Generations 5, 6, and 7 of lines selected for multitrait immune response Selection direction2 1

Interaction

Replicate

Trait

Generation

H

L

P

1

2

P

P

Index

5 6 7 5 6 7 5 6 7 5 6 7 5 6 7

1.18 2.96 3.86 1.31 1.31 1.74 .283 .282 .826 .275 .444 .608 1.10 1.27 1.58

1.12 2.77 3.38 .97 1.03 1.22 .305 .205 .605 .258 .445 .562 1.25 1.26 1.53

NS NS .001 .005 NS .024 NS NS .022 NS NS NS .035 NS NS

1.09 2.69 3.53 1.02 .83 1.22 .258 .078 .673 .274 .463 .587 1.17 1.32 1.61

1.21 3.04 3.71 1.26 1.50 1.75 .329 .410 .758 .259 .426 .582 1.18 1.21 1.51

.027 .020 NS .034 .017 .022 NS .0007 NS NS NS NS NS .014 .020

.0008 NS .093 .011 NS NS NS NS .008 NS .032 NS NS NS NS

MG PM CCA PHA

x

Traits are: Index = multitrait immune response value; MG = Mycoplasma gallisepticum antibody; PM Pasteurella multocida antibody; CCA = carbon clearance response; PHA = wing web response. 2 H = high immunoresponsiveness; L = low immunoresponsiveness.

Downloaded from http://ps.oxfordjournals.org/ at Kainan University on April 9, 2015

Index weights used for the fifth, sixth, and seventh generations, respectively, were: .379, .522, and .512 for MG; .335, .546, and .398 for PM; .335, 1.52, and 1.36 for CCA; and .409, 1.14, and 1.14 for PHA. Heritability estimates of the four individual traits, MG, PM, CCA, and PHA and of the index were estimated by using three different methods. Table 2 presents Restricted Maximum Likelihood (REML) estimates of heritabilities from sire variance components of Generations 5, 6, and 7 and a mean (weighted by number of birds per generation) of the three generation means. Heritabilities ranged from 0 to .43 in individual generations and from .06 to .31 when averaged across generations. Estimates of heritabilities obtained by using parent-offspring correlations in the two cycles of selection and the weighted means of these estimates and from chickens pooled by selection direction are presented in Table 3. Heritability estimates

ranged from -.13 to .16 in individual cycles of selection and from 0 to .12 when averaged across selection cycles. Estimates of realized heritabilities of the four traits and of the index are in Table 4. Again, estimates from each cycle of selection and a weighted mean are shown. These estimates ranged from -2.42 to 2.18 in individual cycles and from -1.21 to 1.52 when averaged across cycles. Correlations, both genetic and phenotypic, between the four traits and between the individual traits and the index were estimated by using two different methods. Phenotypic and genetic correlations from REML estimates of variance and covariance are reported in Table 2. Phenotypic correlations among individual immuneresponse traits ranged from -.15 to .31 in individual generations and from -.06 to .12 when pooled across generations. Phenotypic correlations pooled across generations between the index and individual traits ranged from .21 between the index and CCA to .60 between MG and the index. Estimates of genetic correlations among immune response traits ranged from -.34 to .61. When pooled across generations, estimates were from -.34 to .58.

25

EFFECT OF SELECTION FOR MULTTTRAIT IMMUNOCOMPETENCE 1

TABLE 2. Heritabilities and genetic and phenotypic correlations, calculated from sire variance and covariance components, among four immune response traits2 and an index of the traits in three generations of lines of divergently selected chickens Trait Generation

Index

Index

5 6 7 5c* 5 6 7 5c 5 6 7 5 6 7 5c

0 .24 .26 .16 .19 .85 .77 .60 .15 .25 .55 .31 .03 .28 .32 .21

5 6 7 5c

.10 .34 .39 .27

MG

PM

X

CCA

PHA

MG

PM

CCA

PHA

,3

.95 1.17 1.06 .12 .43 .39 .31 .03 .04 .31 .12 .02 .02 -.06 -.01 -.02 -.03 -.07 -.04

.45 .45 .45 .17 .39 .28 .04 .34 .26 .21 .03 -.15 .13 0 -.01 -.15 -.01 -.06

.29 -.19 -.19

.29

.04 .25 .15

-.34

.61 .55 .58 0 .13 .05 .06 0 -.04 -.04 -.03

.33

-.34

.33 .46 .46 .10 .12 .01 .08

heritabilities are on diagonal (in bold print), genetic correlations are above diagonal, and phenotypic correlations are below diagonal. 2 Traits are: Index = multitrait immune response value; MG = Mycoplasma gallisepticum antibody; PM = Pasteurella multocida antibody; CCA = carbon clearance response; PHA = wing web response. 3 Values were not estimable when estimate of sire variance was nearly 0. 4 5c = mean of three generations, weighted by number in each generation.

TABLE 3. Heritabilities, estimated by using parent-offspring correlation, of four immune response traits1 and an index of the traits in three generations of chickens divergently selected for multitrait immune response Trait 2

Generation

Index

MG

PM

CCA

PHA

H

5 to 6 6 to 7 5P

.16 -.05 .05 .04 .03 .03 .10 -.01 .05

.10 .14 .12

5 to 6 to 5c 5 to 6 to 5?

.03 .15 .09 .04 .06 .05 .03 .13 .08

.03 -.13 -.05

L

.06 .09 .07 .13 .03 .08 .10 .07 .09

Pool

Total

6 7 6 7

.01 .10 .04 .03 -.07 -.02

.13 .10 .11 .12 .12 .12

J Traits are: Index = multitrait immune response value; MG = Mycoplasma gallisepticum antibody; PM = Pasteurella multocida antibody; CCA = carbon clearance response; PHA = wing web response. 2 Pools are: H = High immune response lines combined; L = low immune response lines combined; Total = All lines combined. ^ = mean of two cycles, weighted by number per cycle.

Downloaded from http://ps.oxfordjournals.org/ at Kainan University on April 9, 2015

Trait

26

KEAN ET AL.

TABLE 4. Realized heritabilities of four immune response traits and an index of the traits in three generations of chickens divergently selected for multitrait immune response Trait1 Replicate

Generation 5 to 6 to 5? 5 to 6 to 5c"

Index

MG

PM

CCA

6 7

.23 .15 .19

.14

6 7

-.01

-.23

.24 .11

-.01

.27 .61 .44 .06 .59 .32

2.18

-.02

.06 .21

.85 1.52 -2.42

0 -1.21

PHA

.84 .11 .48 .10 .21 .15

J

Traits are: Index = multitrait immune response value; MG = Mycoplasma gallisepticum antibody; PM Pasteurella multocida antibody; CCA = carbon clearance response; PHA = wing web response. zjc = mean of two cycles, weighted by number per cycle.

pooled. Genetic correlations between the index and individual traits ranged from -.42 to .96 when the two cycles were pooled. Some differences were noted between high-response lines and low-response lines and between Replicates 1 and 2 with respect to juvenile body weights (Table 7). In Generation 5, chicks in the low lines were heavier than chicks in the high lines at 2 wk of age. In Generation 6, the low lines were heavier at 6 wk and at 12 wk of

TABLE 5. Parent-offspring correlations among four immune response traits and an index of the traits in three generations of lines of divergently selected chickens Progeny trait1 Index MG

Parental trait2 Generation 5 to 6 to 5? 5 to 6 to

6 7

CCA PHA

5 to 6 to 5c 5 to 6 to 5c 5 to 6 to 5c

6 7 6 7 6 7

PMG .03 .08 .06

6 7

X"

PM

PIndex

.07 .12 .10 -.10 -.01 -.06 .05 -.03 .01 .12 .02 .07

-.06 .06 0 .05 -.03 .01 0 -.05 -.03

PPM .06 .01 .04 .06 .04 .05

-.03 -.02 -.03 .02 .04 .03

PCCA

PPHA

-.02 -.04 -.03 -.07 0 -.04 -.04 -.11 -.08

.05 .01 .03 .03 -.05 -.01 -.08 -.04 -.06 .02 .01 .02

.03 .01 .02

1 Traits are; Index = multitrait immune response value; MG = Mycoplasma gallisepticum antibody; PM = Pasteurella multocida antibody; CCA = carbon clearance response; PHA = wing web response. 2 Traits are: PIndex = multitrait immune response value of midparent; PMG = Mycoplasma gallisepticum antibody response of midparent; PPM = Pasteurella multocida antibody response of midparent; PCCA = carbon clearance response of midparent; PPHA = wing web response of midparent. 35c" = mean of two cycles, weighted by number per cycle.

Downloaded from http://ps.oxfordjournals.org/ at Kainan University on April 9, 2015

Parent-offspring correlations are reported separately for trait x in parents with trait y in offspring and trait y in parents with trait x in offspring (Table 5). These correlations ranged from -.11 to .06 between traits and from -.10 to .12 between the index and individual traits. Estimates of genetic correlations, estimated from parent-offspring correlations (Table 6), ranged from -.66 to .21 for individual traits in a single cycle and from -.66 to .05 when the two cycles were

27

EFFECT OF SELECTION FOR MULTTTRAIT IMMUNOCOMPETENCE

TABLE 6. Genetic correlations, calculated from parent-offspring correlations, among four immuneresponse traits and an index of the traits in three generations of lines of divergently selected chickens Trait 1

Trait

Generation

Index

5 to 6 6 to 7

MG PM

5 to 6 to X 5 to 6 to

MG

PM

CCA

PHA

.88 1.04 .96

-.42

.15

.81 .12 .47 .21 -.41 -.10 -.54

,2

6 7

-.42 -.02

.15 -.13

-.02

-.13 -.66

6 7

5 to 6 6 to 7 X

.05

'Traits are: Index = multitrait immune response value; MG = Mycoplasma gallisepticum antibody; PM = Pasteurella multocida antibody; CCA = carbon clearance response; PHA = wing web response. Correlations were not estimable when sire variance estimates were zero. ^ = mean of two cycles, weighted by number per cycle.

age. In Generation 6, chicks in Replicate 2 were significantly heavier than those in Replicate 1 at 2 wk of age but at 2 wk of age the ranking was reversed. In Generation 7, Replicate 1 chicks weighed more at 12 wk of age. All these differences,

however, had disappeared by 20 wk of age, and no differences between lines were noted in adult body weights. Egg production records from Generations 5 and 6 are described in Table 8. No differences (P > .05) were noted between

TABLE 7. Least squares means of body weights by selection direction and replicate, and levels of significance of Generations 5, 6, and 7 of lines selected for multitrait immune response

BW

Generation

H

L

P

1

•(g)

2 wk

5 116 122 .01 6 129 132 NS 7 125 130 NS 6 wk 5 410 419 NS 6 425 443 .01 7 399 412 NS 12 wk 5 877 892 NS 6 838 893 .08 7 873 912 NS 20 wk 5 1,110 1,130 NS 6 1,392 1374 NS 7 1,390 1,397 NS 32 wk 5 1,766 1,740 NS 1,781 6 1,740 NS l 7 51 wk 5 1,916 NS 1,975 6 1,964 1,968 NS 7 'Data were not available at the time of this analysis.

Interaction

Replicate

Selection direction

1

2

P

P

•(g)

119 128 129 453 419 412 896 842 917 1,123 1,373 1,420 1,763 1,761

119 133 126 410 416 399 873 890 868 1,117 1,394 1,367 1,742 1,759

NS .02 NS NS .0001 NS NS NS .04 NS NS NS NS NS

NS NS NS .07 NS NS .08 NS NS NS NS NS NS NS

1,977 1,958

1,933 1,954

NS NS

NS NS

Downloaded from http://ps.oxfordjournals.org/ at Kainan University on April 9, 2015

-.54 .05

-.66 CCA

28

KEAN ET AL.

TABLE 8. Least squares means of egg production measures, by selection direction and replicate, and levels of significance of Generations 5 and 6 of lines selected for multitrait immune response Selection direction

Interaction

Replicate

Generation

H

L

P

1

2

P

P

AFE

5 6 5 6 5 6 5 6 5 6 5 6 5 6

167 170 48.1 49.8 .381 .344 .725 .844 .734 .734 .653 .622 .509 .634

169 172 47.7 49.1 .356 .291 .728 .834 .709 .763 .641 .638 .458 .624

NS NS NS NS NS NS NS NS NS NS NS NS NS NS

167 170 47.8 49.7 .366 .328 .716 .847 .703 .759 .653 .619 .480 .636

168 172

NS NS NS NS NS NS NS NS NS NS NS NS NS NS

NS NS NS NS NS NS NS NS NS NS NS NS NS NS

EGGWT PI P2 P3 P4 PT

48.1 49.2 .372 .306 .741 .831 .741 .738 .638 .638 .486 .622

'Traits are: AFE, age in days at first egg; EGGWT, average egg weight in grams at 32 wk of age; PI, egg production frequency from 20 to 28 wk; P2, egg production from 29 to 36 wk; P3, egg production from 37 to 44 wk; P4, egg production from 45 to 52 wk; PT, egg production from 20 to 52 wk.

replicates or between high- and lowresponse lines for the egg production traits measured, which included age at first egg, average egg weight at 32 wk of age, four 8-wk production periods starting at 20 wk of age, and total egg production in the 32-wk period from 20 to 52 wk of age. Percentage mortality in each generation is shown in Table 9. Significant interactions were present because Line 1L had significantly (P < .05) lower juvenile mortality (0 to 20 wk) than did Line IH in Generations 5 and 6 (data not shown). No significant differences were noted in adult mortality in Generations 5 or 6 or in Generation 7 up to 20 wk of age.

DISCUSSION Divergent selection for multitrait immune response has produced phenotypic differences in immune response among the lines in this study. In Generation 7, lines selected for high immune-response had greater humoral immune response and greater index values than did lines selected for low immune-response. Because the selection has been on an index composed of all four traits, intensity of selection for each individual trait has been decreased. It may require several more generations before significant differences are realized among these lines for all of the immune-response traits. Estimates of

TABLE 9. Juvenile and adult mortality by selection direction and replicate, and levels of significance of Generations 5, 6, and 7 of lines selected for multitrait immune response Selection direction Mortality

Generation

0 to 20 wk

5 6 7 5 6

H

L

P

1

NS NS NS NS NS

7.6 5.3 8.0 12.0 6.0

— (%)

20 to 52 wk

9.3 7.0 8.0 11.0 6.0

7.3 4.0 6.0 10.0 8.0

Interaction

Replicate 2

P

P

NS NS NS NS NS

<.02 <.001 NS NS NS

(%) 9.0 5.7 6.0 9.0 8.0

Downloaded from http://ps.oxfordjournals.org/ at Kainan University on April 9, 2015

Trait1

EFFECT OF SELECTION FOR MULTITRAIT IMMUNOCOMPETENCE

of the low selection pressure and small changes among lines, correction for bias was not included in this analysis. Phenotypic correlations among the four traits of immunoresponsiveness were generally small (Tables 2 and 5), with the exception of the correlation between MG and PM. Positive, but often small, correlations between measures of antibody production have been found in other experiments conducted with pigs (Buschmann et al, 1985) and with chickens (Cheng and Lamont, 1988; Heller et al, 1992). Mallard et al. (1992) reported a very small positive and nonsignificant correlation of .01 in males and .05 in females. After one generation of selection, using EBVs that included response to HEWL but did not include response to TGAL, the high line exhibited a significantly greater response to TGAL than did the low line. Antibody response to SRBC, another unselected trait, was also greater in the high line than in the low line. Several of the other phenotypic correlations in this study, although small, were negative. Biozzi et al. (1979) also reported some negative correlations between measures of different facets of immune response, especially between phagocytic ability and antibody production, but found no significant correlation between either of these two traits and measures of cellular immune response. Mallard et al. (1992) reported a greater delayed-type hypersensitivity response to dinitrochlorobenzene (DNCB), an unselected trait, in the line of pigs selected for low multitrait immune responsiveness than in the corresponding line selected for high response. Estimates of genetic correlation between immune response traits (Tables 2 and 6) from the present study were generally negative. These may slow further progress in this selection experiment. Improved accuracy, obtained by testing more generations, would seem to be necessary before these estimates should be included in index calculations. More information concerning the true economic values of these immuneresponse traits is needed as well. Hazel and Lush (1942) point out that the first important part of using a selection index is to have the proper economic weights.

Downloaded from http://ps.oxfordjournals.org/ at Kainan University on April 9, 2015

heritabilities, although low, predict that successful long-term selection is feasible. Several of the estimates of heritability (ranging from 0 to .43 as estimated by using variance components or parentoffspring correlation) are similar to those found in other experiments involving traits of immune response. Biozzi et al. (1979) reported a heritability of .18 for antibody response to SRBC in mice. Eide et al. (1991) reported realized heritability of .19 for antibody response to diphtheria toxoid in goats. After one generation of selection, Mallard et al. (1992) characterized several immune response traits in swine and estimated heritabilities ranging from 0 to .25. In chickens, van der Zijpp et al. (1983) found realized heritability of .20 when selecting for increased antibody titer to SRBC and .52 when selecting for decreased titers. Martin et al. (1990) reported realized heritabilities of .23 and .25 in Generations 10 to 14 of selection for high and low antibody production to SRBC, respectively. Realized heritabilities of Leitner et al. (1992) for antibody response to Escherichia coli were similar, with estimates of .23 for high response and .32 for low response. The estimates of realized heritabilities varied widely in the generations evaluated in the present study. These fluctuations may be influenced by many factors. Because the selection was conducted on the index, the actual selection differentials for individual traits were very small in some instances (data not shown). Additionally, because no control group was kept, responses in each direction of selection were assumed to be equal. This has not been confirmed and, based upon recent studies by Pinard et al. (1992), may possibly be unequal. Additionally, incidence of nonresponders, as has been noted in other immune response selection experiments (e.g., Pitcovski et al, 1987), has been low (data not shown). This implies that the low-immune-response line has not been selected to a point below which no response can be measured. The estimates of heritability may also have been biased because selection has occurred (Falconer, 1989). The selection pressure on each individual trait has not been overly intense, however, and because

29

30

KEAN ET AL.

sponse in poultry demonstrate that, with simultaneous selection, one can successfully alter humoral immune response without compromising cell-mediated immunity, reticuloendothelial clearance, and growth and egg production traits. Further tests, including direct challenge with specific pathogens to determine effects of altered immunophysiology on resistance traits, will elucidate the effects of multitrait immune-response selection on disease resistance. ACKNOWLEDGMENTS

The authors thank the workers at the Iowa State University Poultry Research Center for excellent animal care and record collection, numerous students and staff for assistance in conducting the assays as well as for expert advice, and Agriculture Canada, Centre for Food and Animal Research, Ottawa, ON, Canada K1A 0C6 (R. W. Fairfull), for generously donating eggs to produce the base population. REFERENCES Bacon, L. D., 1992. Measurement of immune competence in chickens. Poult. Sci. Rev. 4:187-195. Biozzi, G., D. Mouton, O. A. Sant Anna, H. C. Passons, M. Gennari, M. H. Reis, V.C.A. Ferreira, A. M. Heumann, Y. Boothillier, O. M. Ibanez, C. Stiffel, and M. Siqueria, 1979. Genetics of immunoresponsiveness to natural antigens in the mouse. Curr. Top. Microbiol. Immunol. 85:31-98. Buschmann, H., H. Krausslick, H. Herrmann, J. Meyer, and A. Kleinschmidt, 1985. Quantitative immunological parameters in pigs—experiences with evaluation of an immunocompetence profile. Z. Tierz. Zuchtungsbiol. 102:189-199. Cheng, S., 1990. Genetic Selection for Immunocompetence in Chickens. Ph.D. dissertation, Iowa State University, Ames, IA. Ann Arbor Microfilm Order No. 91-01338. Cheng, S., and S. J. Lamont, 1988. Genetic analysis of immunocompetence measures in a White Leghorn Chicken line. Poultry Sci. 67:989-995. Cheng, S., M. F. Rothschild, and S. J. Lamont, 1991. Estimates of quantitative genetic parameters of immunological traits in the chicken. Poultry Sci. 70:2023-2027. Detilleux, J. C, A. E. Freeman, and L. D. Miller, 1991. Comparison of natural transmission of bovine leukemia virus in Holstein cows of two genetic lines selected for high and average milk production. Am. J. Vet. Res. 52:1551-1555.

Downloaded from http://ps.oxfordjournals.org/ at Kainan University on April 9, 2015

Although this index was designed with the expectation that the weights (the relative amount of value given to responses to each of the four measures of immunocompetence) used would balance selection pressure evenly across the four traits, the correlations between the index and individual traits calculated from Generations 5 to 7 suggest that this may not have been effective. The correlations, both genetic and phenotypic, between the index and MG are consistently larger than those between the index and any of the other three traits (Tables 2 and 5). Gains in selection have also been more striking in MG, although this gain is probably also due to a higher heritability. In Generations 6 and 7, the economic weights (as in Equation 1) for MG and PM were divided in half to increase pressure on CCA and PHA. Index values still were more strongly correlated with MG (Tables 2 and 5) than with CCA or PHA but improvements were made in correlations between the index and CCA and the index and PHA in Generations 6 and 7. The minimal differences among lines in measures of production traits may be a result of a lack of correlation between these traits and the immune-response traits included in this study or of low selection intensity. Negative correlations have been reported between body weight and antibody production to SRBC in other selection experiments (Siegel et al., 1982; van der Zijpp, 1983). In another study, van der Zijpp (1984) found a small, positive correlation between antibody production to Newcastle's disease vaccine and body weight gain. In Generations 5 and 6 in this experiment, Line 1L had lower mortality from 0 to 20 wk than did Lines 1H or 2L, causing a significant interaction. In recent studies (Nelson and Lamont, unpublished data) low-immune-response lines show greater resistance to Marek's disease than do high-immune-response lines, suggesting that specific resistance differences may contribute to observed differences in general mortality. In summary, the results of this longterm, multitrait selection for immune re-

EFFECT OF SELECTION FOR MULTTIRAIT IMMUNOCOMPETENT

Leitner, G., Z. Uni, A. Cahaner, M. Gutman, and E. D. Heller, 1992. Replicated divergent selection of broiler chickens for high or low early antibody response to Escherichia coli vaccination. Poultry Sci. 71:27-37. Mallard, B. A., B. N. WUkie, B. W. Kennedy, and M. Quinton, 1992. Use of estimated breeding values in a selection index to breed Yorkshire pigs for high and low immune and innate resistance factors. Anim. Biotech. 3:257-280. Martin, A., E. A. Dunnington, W. B. Gross, W. E. Briles, R. W. Briles, and P. B. Siegel, 1990. Production traits and alloantigen systems in lines of chickens selected for high or low antibody responses to sheep erythrocytes. Poultry Sci. 69:871-878. Meeker, D. L., M. F. Rothschild, L. L. Christian, C. M. Warner, and H. T. Hill, 1987. Genetic control of immune response to pseudorabies and atrophic rhinitis vaccines: I. Heterosis, general combining ability and relationship to growth and backfat. J. Anim. Sci. 64:407-413. National Research Council, 1984. Nutrient Requirements of Poultry. 8th rev. ed. National Academy Press, Washington, DC. Pevzner, I. Y., I. Kujdych, and A. W. Nordskog, 1981a. Immune response and disease resistance in chickens, n. Marek's disease and immune response to GAT. Poultry Sci. 60:927-932. Pevzner, I. Y., H. A. Stone, and A. W. Nordskog, 1981b. Immune response and disease resistance in chickens I. Selection for high and low titer to Salmonella pullorum antigen. Poultry Sci. 60: 920-926. Pinard, M.-H., J.A.M. van Arendonk, M.G.B. Nieuwland, and A. J. van der Zijpp, 1992. Divergent selection for immune responsiveness in chickens: Estimation of realized heritability with an animal model. J. Anim. Sci. 70: 2986-2993. Pitcovski, J., E. D. Heller, A. Cahaner, and B. A. Peleg, 1987. Selection for early responsiveness of chicks to Escherichia coli and Newcastle disease virus. Poultry Sci. 66:1276-1282. SAS Institute, 1985. SAS® User's Guide: Statistics. Version 5 Edition. SAS Institute Inc., Cary, NC. Siegel, P. B., and W. B. Gross, 1980. Production and persistence of antibodies in chickens to sheep erythrocytes. 1. Directional selection. Poultry Sci. 59:1-5. Siegel, P. B., W. B. Gross, and J. A. Cherry, 1982. Correlated responses of chickens to selection for production of antibodies to sheep erythrocytes. Anim. Blood Groups Biochem. Genet. 13: 291-297. Snedecor, G. W., and W. G. Cochran, 1989. Statistical Methods. 7th ed. The Iowa State University Press, Ames, IA. van der Zijpp, A. J., 1983. Breeding for immune responsiveness and disease resistance. World's Poult. Sci. J. 39:118-131. van der Zijpp, A. J., 1984. Breeding for immune responsiveness and disease resistance in poultry. Pages 1-11 in: Proceedings 33rd Annual National Poultry Breeders Roundtable, St. Louis,

Downloaded from http://ps.oxfordjournals.org/ at Kainan University on April 9, 2015

Eide, D. M., T. Adnoy, G. Klemetsdal, L. L. Nesse, and H. J. Larsen, 1991. Selection for immune response in goats: The antibody response to diphtheria toxoid after 12 years of selection. J. Anim. Sci. 69:3967-3976. Falconer, D. S., 1989. Introduction to Quantitative Genetics. 3rd ed. John Wiley and Sons, Inc., New York, NY. Gavora, J. S., and J. L. Spencer, 1978. Breeding for genetic resistance to disease: specific or general? World's Poult. Sci. J. 34:137-148. Gavora, J. S., and J. L. Spencer, 1983. Breeding for immune responsiveness and disease resistance. Anim. Blood Groups Biochem. Genet. 14: 159-180. Glick, B., K. Sato, and F. Cohenour, 1964. Comparison of the phagocytic ability of normal and bursectomized birds. J. Reticuloendothel. Soc. 1: 442-449. Gowe, R. S., and R. W. Fairfull, 1980. Performance of six long-term multi-trait selected Leghorn strains and three control strains, and a strain cross evaluation of the selected strains. Pages 141-162 in: Proceedings of the South Pacific Poultry Science Convention, Auckland, New Zealand. Gross, W. B., P. B. Siegel, R. W. Hall, C. H. Domermuth, and R. T. DuBoise, 1980. Production and persistence of antibodies in chickens to sheep erythrocytes. 2. Resistance to infectious diseases. Poultry Sci. 59:205-210. Hazel, L. N., 1943. The genetic basis for constructing selection indexes. Genetics 28:476-490. Hazel, L. N., and J. L. Lush, 1942. The efficiency of three methods of selection. J. Hered. 33:393-399. Heller, E. D., G. Leitner, A. Friedman, Z. Uni, M. Gutman, and A. Cahaner, 1992. Immunological parameters in meat-type chicken lines divergently selected by antibody response to Escherichia coli vaccination. Vet. Immunol. Immunopathol. 34:159-172. Hill, W. G., 1972. Estimation of realized heritabilities from selection experiments. I-Divergent Selection. Biometrics 28:747-766. Karakoz, I., J. Krejci, K. Hala, B. Blaszczyk, T. Hraba, and J. Pekarek, 1974. Genetic determination of tuberculin hypersensitivity in chicken inbred lines. Eur. J. Immunol. 4:545-548. Kehrli, M. E., Jr., K. A. Weigel, A. E. Freeman, J. R. Thurston, and D. H. Kelley, 1991. Bovine sire effects on daughters' in vitro blood neutrophil functions, lymphocyte blastogenesis, serum complement, and conglutinin levels. Vet. Immunol. Immunopathol. 27:303-319. Lamont, S. J., 1986. Genetic associations of reticuloendothelial activity in chickens. Vol. XI. Pages 643-647 in: Proceedings of Third World Congress of Genetics Applied to Livestock Production. G. E. Dickerson and R. K. Johnson, ed. University of Nebraska, Lincoln, NE. Umont, S. J., and J. R. Smyth, Jr., 1984. Effect of selection for delayed amelanosis on immune response in chickens. 2. Cell-mediated immunity. Poultry Sci. 63:440-442.

31

32

KEAN ET AL.

MO. van der Zijpp, A. J., K. Frankena, J. Boneschanscher, and M.G.B. Nieuwland, 1983. Genetic analysis of primary and secondary immune response in chickens. Poultry Sci. 62:565-572. Warner, C. M., D. L. Meeker, and M. F. Rothschild, 1987. Genetic control of immune responsive-

ness: A review of its use as a tool for selection for disease resistance. J. Anim. Sci. 64:394-406. Weigel, K. A., A. E. Freeman, M. E. Kehrli, Jr., J. R. Thurston, and D. H. Kelley, 1992. Relationship of in vitro immune function with health and production in Holstein cattle. J. Dairy Sci. 75: 1672-1679.

Downloaded from http://ps.oxfordjournals.org/ at Kainan University on April 9, 2015