GENETIC AND MATERNAL EFFECTS ON PIG WEIGHTS, GROWTH AND PROBE BACKFAT IN DIALLEL CROSSES OF HIGH- AND LOW-FAT LINES OF SWINE 1

Transcription

1 GENETC AND ATERNAL EFFECTS ON PG WEGHTS, GROWTH AND PROBE BACKFAT N DALLEL CROSSES OF HGH- AND LOW-FAT LNES OF SWNE 1 B. Bereskin 2 and H. O. Hetzer a U.S. Department of Agriculture 4 Beltsville, D 275 ABSTRACT Two Duroc and two Yorkshire lines of pigs that had been selected at Beltsville Agricultural Research Center for 12 and 1 generations, respectively, for either thinner or thicker backfat were mated to produce all possible pure lines and reciprocal crosses in 1967, 1969 and 197. Data for littermate gilts and barrows from 136 litters were analyzed to estimate genetic and maternal influence on individual pig weights at birth, 21 d, 56 d and 14 d of age; age at 79.4 kg; average backfat thickness at 79.4 kg and postweaning average daily gain (56 d to 79.4 kg). Pure-line gilts differed among breed-lines (P.5 or P.O1) for all traits except weight at 56 d. Gilts of the two low-far lines were heavier than gilts of the two high-fat lines through 56 d of age, but Yorkshire low-fat gilts were iightest at 14 d, were oldest at 79.4 kg and had the slowest daily gain, in addition to the least backfar. The Duroc low-fat line gilts were heaviest at 14 d, youngest at 79.4 kg and were second thinnest in backfat. Among pure-line barrows, the low-fat lines were heaviest at birth, at 21 d and at 14 d and were thinnest in backfat. Line-cross gilts were heavier than pureqine gilts at all four ages, were younger at 79.4 kg and higher in daily gain. Among barrows, line crosses were heavier in all weights except at 21 d, were younger at 79.4 kg and were higher in daily gain than pure lines. Differences between pure lines and line crosses in backfat were not significant for either sex. Heterosis varied from 6.5 to 16.7% among weights and growth traits. Pigs of both sexes differed among breed-lines in general combining ability for all traits except 21-d weight, and differed in maternal ability for weights through 56 d and for backfat. Specific combining ability (SCA) was significant only for intra-breed crosses for weight at 21 d, and for inter-breed, intraqine crosses for 21- and 56-d weights and for age at 79.4 kg among gilts, with no significant effects in SCA for any trait among barrows. General combining ability was not correlated with maternal effects for any trait except 21-d weight, for which they were positively correlated (r>.8). (Key Words: Pigs, Performance Traits, Diallel Crossing, Heterosis, Genetic Parameters, aternal Effects.) 1The authors acknowledge with appreciation the contributions of W. H. Peters, Head, Animal Operations, for managing the swine herd and recording the necessary performance data during this experiment; the contributions of Bonnie organ, Statistical Assistant; and the comments by Dr. D. L. Kuhlers, Auburn Univ., and by Dr. L. D. Young, Roman L. Hruska U.S. ARC, Clay Center, NE. 2USDA, Agr. Res. Sew., Beltsville Agr. Res. Center, Animal Sci. nst., Nonruminant Anita. Nutr. Lab., Beltsville, D 275. a USDA, Agr. Res. Serv., Beltsville Agr. Res. Center, Beltsville, D 275, retired. 4ention of trade name, proprietary product or vendor does not constitute a guarantee or warranty of the product by USDA or imply its approval to the exclusion of other products or vendors that may also be suitable. Received November 5, Accepted arch 18, ntroduction With general use of breed and strain crossing in current swine production, it is important to estimate basic parameter values with which to evaluate and identify superior crossing combinations. These parameters include average heterosis, general and specific combining ability and breed maternal and maternal heterosis effects. Henderson (1948) first formulated comprehensive least-squares procedures to analyze single crosses of inbred lines of swine in order to estimate various sets of these parameters. Hetzer et al. (1961) adapted those methods in analyzing single crosses among six inbred lines of swine. Harvey (196) extended Henderson's methods to include disproportionate subclass numbers and estimates of 395 J. Anim. Sci :395-48

2 396 BERESKN AND HETER pure-line performance and mean heterosis effects in analyses of diallel crosses. Bereskin et al. (1974) used methods developed by Harvey (196) to analyze a diallel experiment involving litter traits in four lines of swine. Schneider et al. (1982) used a complex split-plot design, while Jungst and Kuhlers (1984) used multiple regression methods to analyze performance of various breeds and breed crosses in order to estimate sets of genetic parameters for swine. The present study was intended to provide simultaneous estimates of average heterosis, pure-line effects, general and specific combining ability and breed direct maternal effects from an analysis of pig weights, growth rate and probe backfat data in a diallel experiment involving four lines, with application to swine production practices. aterials and ethods Two Duroc (D) and two Yorkshire (Y) lines of swine were selected at the USDA Beltsville Agricultural Research Center beginning in 1954 and 1956, respectively, for either greater (DH, YH) or lesser (DL, YL) backfat (Hetzer and Harvey, 1967). These four lines were used to produce all possible reciprocal crosses and pure lines that farrowed in the spring of 1967, 1969 and 197. Alternative management priorities precluded use of facilities in Available females were randomly assigned to the 16 possible subclasses. Only second litters out of dams 18 mo of age were represented in this study. The plan in each of the 3 yr was to mate up to four randomly selected active boars in each of the four lines to sows of the four lines. Different boars were used in each year. A total of 39 boars sired progeny included in these analyses. Nearly all of the sires produced progeny in two or more lines of dams. As a result, the residual variance in the analyses consisted essentially of the pooled variance among samples of sires within the 16 subclasses. A total of 136 litters were produced in the 3 yr, each with at least one gilt and one barrow selected at random from among healthy pigs weaned at 56 d (table 1). n litters with more than one gilt and(or) more than one barrow tested, one gilt and one barrow were selected at random for inclusion in this analysis. Pigs were tested from weaning at 56 d of age to 14 d of age, or to 79.4 kg, whichever was later. Traits analyzed were individual pig weights at birth, at 21 (W21), 56 (W56) and 14 (W14) d of age, age in days at 79.4 kg (AGE), average daily gain from 56 d of age to 79.4 kg (ADG) and average backfat thickness (ABF), based on probes at the first and last rib and last lumbar vertabra 6 cm on each side of the mid-line of the back at 79.4 kg. The lean meter (Andrews and Whaley, 1955) was used in making the probes. nbreeding was assumed to be for litters (pigs) from inter-breed crosses and 3 to 5% for litters from intra-breed line crosses. nbreeding of pure-line pigs averaged about 29% for Durocs and 27% for Yorkshires. nbreeding of all dams of both pure-line and line-cross litters ranged from 25 to 29%. odel (1). Henderson (1948) first formulated and Harvey (196) extended statistical methods for analyzing line-cross or diallel data involving multiple classifications and disproportionate subclass numbers. Harvey's methods were largely applied in the present analyses. Exact procedures for setting up equations in the TABLE 1. DSTRBUTON OF LTTERS REPRESENTED BY BREED AND LNE OF SRE AND DA a Breed and line Breed and line of sire of dam D-High D-Low Y-High Y-Low Totals D-high D-low Y-high Y-low Totals ad-high = Duroc line selected for high backfat thickness. D-low = Duroc line selected for low backfat thickness. Y-high = Yorkshire line selected for high backfat thickness. Y-low = Yorkshire line selected for low backfat thickness.

3 DALLEL CROSSES OF HGH- AND LOW-FAT SWNE 397 least-squares (LS) analyses were those by Bereskin and Norton (1982). The primary biometric model used was similar to model (1) in Bereskin et al. (1974). Pure-line effects were estimated only from the four pure lines, while general (GCA) and specific (SCA) combining ability, breed-line maternal effects and specific reciprocal effects were estimated from line crosses only. n addition, the models included effects of average heterosis from comparisons of pure-line and line-cross LS means, year of farrow, period in year farrowed, interaction of year period and one or more continuous variables as covariates. The covariates were litter size farrowed (live + stillborn, No), pig weight at 56 d and inbreeding percentages of the dam (F D ) and pig (Fp) in various preliminary or final analyses. Two periods in year farrowed included pigs born either before or after arch 2. The spring farrowing period usually extended from about February 15 until ay 1. Weather conditions generally differed in the two periods. Also, arch 2 was about the mid-point in numbers of litters farrowed. Analyses were computed separately with the gilt and barrow data, and for each littermate pair expressed as gilt - (minus) barrow in order to assess any sex differences for the various effects and traits. The preceeding models expressed performance of pigs and lines in terms of particular genetic and maternal effects. odel (2). n complementary analyses, the same data sets were analyzed by modifying model (1) into model (2) as follows: The three degrees of freedom (df) for pure lines provided comparisons between the Duroc and Yorkshire breeds, between lines selected for high or low backfat and for the interaction of breeds lines. The 11 df among line crosses were divided into the three sets of four crosses with three df within each of the three sets. These sets correspond to SCA effects in model (1): intra-breed; inter-breed, intraqine; and inter-breed, inter-line. ntra-breed line crosses included DH DL, YH YL and their reciprocals. nter-breed, intra-line line crosses were DH YH, DL YL and their reciprocals. nter-breed, inter-line line crosses included DH YL, DL YH and their reciprocals. The three df within each set included two df for comparisons between reciprocal crosses and one df to compare the pairs of reciprocal crosses. The latter comparison in each set corresponds to one of the comparis.ons among pure lines: between breeds, between lines or their interaction. Each of the preceeding comparisons among pure lines and among line crosses reflects differences in GCA and in maternal effects. Also, SCA was not a factor in the preceeding comparisons among line crosses. However, the remaining two df among line crosses were used to estimate SCA effects in the same manner as for model (1). The other effects included in model (1) were also fitted here with model (2). The analyses with model (2) provided further comparisions relative to the effects of interest, not directly available with model (1), without compromising those findings. Remits Preliminary Analyses. odels that included F D or Fp in addition to N o or W56 were fitted to the gilt and barrow data. For gilts, R 2 based on models with F D included differed from models without F D by.1 for pig weight at birth and only in the third decimal, at most, for the other traits. A minor difference in the significance levels for maternal effects also was noted with models with or without F D. For barrows, R 2 based on models with or without F D differed only in the third decimal at most. Only among pure lines were minor differences in significance levels noted for models with or without F D. With Fp in the model, no differences in significance levels were noted for any trait except for mean heterosis effects in gilt data. With Fp, no trait exhibited significant mean heterosis effects, compared with significant levels for all traits except ABF, with Fp deleted. Among barrows, with Fp in the model, no significant mean heterosis effect was noted for any trait. With Fp deleted, significant heterosis was noted for all traits except W21 and ABF. With both gilts and barrows, R 2 values for models with or without Fp differed by 1% or less for all traits. This result would appear to confirm the basic assumption that heterosis in crosses of partially inbred lines consists largely of the recovery from accumulated effect of reduced heterozygosity or random drift in gene frequencies due to limited population size, i.e., inbreeding depression, in the parental stock (Dickerson, 1972). According to Eisen et al. (1983), specific reciprocal effects do not contain sexqinked effects, as previously assumed (Henderson,

4 ]98 BERESKN AND HETER 1948; Harvey, 196), but may represent specific cytoplasmic effects and possibly differences in gene frequency between reciprocal crosses. However, their exact genetic meaning remains uncertain. n any case, variance among estimated specific reciprocal effects did not approach statistical significance for any trait in preliminary analyses of either the gilt or z barrow data sets. As a result, subsequent final analyses excluded reciprocal effects along with F D and Fp, but included N o as a continuous covariate in models for all traits except ADG, which had W56 as a continuous covariate. Also, the final model included effects of years, periods within year and year period inter- ~, actions, in addition to effects of average heterosis, pure lines, maternal effects, GCA and SCA and is referred to as model (1). All effects e~ except residual were assumed to be fixed, whereas residual effects were assumed to be random and normally 2 and independently distributed (ND; O, OE) for use in tests of statistical significance. Analyses voith odel (1) ncluded in table 2 are overall LS means, LS constants, F-ratios and indications of statistical significance, plus related statistics from analysis of the gilt data. ncluded in table 3 are similar e~ statistics from analysis of the littermate barrow ~: data. ncluded in table 4 are results from E- analysis of the gilt - barrow data with model (1), to evaluate interaction effects with sex of pig. Effects of years, periods within year and year x period interactions are not presented because they were not pertinent to the main results. Based on overall LS differences (table 4), barrows averaged heavier than gilts in all. weights, younger in AGE, faster in ADG and e~ thicker in ABF. These differences were statistic- o ally significant for all traits except birth weight o~ and W21. > ean Heterosis. Least-squares constants for ~" type of mating (tables 2 and 3) show that z among gilts, line crosses were heavier than ~,i pure lines for all four weights, were younger in AGE and had higher ADG (P.5 or P.1). Among barrows, line crosses exceeded pure ~" lines in weight at birth (P.1), at 56 and 14 d, were younger in AGE and had higher ADG (P.1). Differences were not significant for ABF among either gilts or barrows. Percent heterosis was slightly higher among gilts than among barrows, with values increasing with age.. ~.h~.. 4,~ ~ o'~ ~;.~ ~ o O~-~ ~'~ ~,~ ~ =.: ~ ~. ~. ~. o. ~ ~

5 DALLEL CROSSES OF' HGH- AND LOW-FAT SWNE 399 oq. q q q qq,q. 9 " " " ~ " " " " qq~q ~ q ~ ~qqqq ~ ~- ~. /~-~' ~1 i ~, ~'~ o "~_o~.~ ~ o.~ ~ 9 ~.~.~.~ ~ ~,~ ~ ~ o o-~-~ ~ "~ ~ ~ ~ ~ "'~ ~.9.9 ~... ~ ~ ~ ~ ~ Ox t~ -V- x; of pig. Essentially no heterosis was noted for ABF. From analysis of gilt - barrow records (table 4), only for AGE was the gilt - barrow difference between pure lines and line crosses significant, indicating an interaction effect9 Pure lines exceeded line crosses in AGE by 2.9 d more (P.5) in gilts than in barrows. Pure Lines9 Pure lines of gilts differed (P.5 or P.1) for au traits except W56 (table 2). The two low-fat lines were heavier than the high-fat lines through W56, but Yorkshire low-fat (YL) gilts were lightest at W14, oldest in AGE, had the smallest ADG among pure lines and were thinnest in ABF, whereas Duroc low-fat (DL) gilts were heaviest at W14, youngest in AGE and had the largest ADG. Among barrows (table 3), the low-fat lines were heaviest at birth, at W21 and W14, were younger in AGE and thinner in ABF, but the difference in ADG was not significant, compared with the high-fat lines9 From the analysis of pure lines with gilt - barrow data (table 4), gilts exceeded barrows in W14 by 5.73 (P.5) and 3.95 (P.lO) kg in the DH and YH lines, respectively, while barrows exceeded gilts by 7.51 kg (P.1) in the YL line. For AGE, gihs were younger than barrows by (P.1) and 9.84 d (P.5) in the DH and YH lines, respectively, but barrows were younger (P.1) than gihs by d in the YL line. Gihs had.34 cm more (P.5) ABF than barrows in the DH line. Gilts also exceeded barrows in ADG in the DH (P.O1) and YH (P.5) lines, while barrows had higher (P.1) ADG by.1 kg/d in the YL line. The F-ratios indicate that the sex effect differed among the pure lines for W14, AGE and ADG (P.1) and for ABF (P.lO). aternal and GCA Effects. aternal and GCA effects were estimated only from line-cross data. Among both gilts and barrows (tables 2 and 3), maternal effects differed (P.O1) among lines for weights at birth and at W21 and W56. For each of these traits, the low-fat lines exceeded the high-fat lines in maternal effects. Also, among both gilts and barrows, maternal effects differed (P.1) for ABF, with the DL line largest for each sex. No significant differences were noted between the sexes in maternal effects among lines for any trait. However, barrows were 1. kg heavier (P.O) than gilts of the YL line for W56. For both gilts and barrows, GCA differed (P.O to P.O1) among lines for all traits

6 4 BERESKN AND HETER e ~ ~ ~ J ~" J + ~ ~ m,, ~ ~ ~ ' "~ ~ ~ ~ " " ~ r~ r~ ~ e +-

7 DALLEL CROSSES OF HGH- AND LOW-FAT SWNE ~.. ~. ~.. ~..... ~ ~ [-, '~". ~ ~ ~,.-~ ~ ~ ~ ~e~,~- O~ ~ ~ ~ e'4 r ~ i [.-. d~ [ ~ m ~ m ~ m ~ [.. ~[ i [ ~ ~ - 9, m~,~.~.~, ~.~.~... ~ ~ ~.~.~.~.~. " ~ = ~.~a.~.a. ~.~. [ ~ 6 #, ## ~o 9 e

8 42 BERESKN AND HETER except W21, with the only difference in significance levels between gilts and barrows being for W56. Lines differed among gilts (P.5) and among barrows (P.1) for W56. With both sexes, the GCA of the DL line was heaviest for weights at all ages, the youngest in AGE, highest in ADG and second thinnest in ABF. With the gilt - barrow data (table 4), gilts exceeded (P.1) barrows in GCA by 8.31 d for AGE in the YL line. Gilts had.35 cm more (P.1) ABF than barrows in GCA for the YH line, but barrows exceeded (P.1) gilts in GCA for ABF by.49 cm in the YL line. Barrows also exceeded (P.5) gilts in GCA for ADG by.5 kg/d in the YL line. However, only for ABF was the interaction effect of sex of pig GCA noted (P.5). Specific Combining Ability (SCA). Heterosis effects generally are separated into mean and specific heterosis. ean heterosis, discussed previously, was measured as the difference between LS estimates of pure line and line cross performance. Specific heterosis, or SCA, of a particular line cross measures the deviation of the mean of the reciprocal crosses of that pair of pure lines from the average performance of all crosses among a group of lines. A significant mean heterosis of SCA effect is assumed to reflect dominance and(or) epistatic gene action (Henderson, 1952). n the present analyses, SCA was based on the performance of three specific groupings of line crosses, with 2 df. ntra-breed line crosses included DH x DL, YH x YL and their reciprocals. nter-breed, intra-line line crosses were DH x YH, DL x YL and their reciprocals. nter-breed, interqine line crosses included DH YL, DL YH and their reciprocals. The sum of the LS constants for the three groups was set equal to O, for computational purposes. The LS constants for SCA among gilts are given in table 2, grouped according to the three types of line crosses, as described previously, with indications of significant differences from zero for individual constants. For all four weights, ADG and AGE, the SCA of inter-breed, intra-line crosses exceeded those of the other types of crosses, but only for W21, W56 and AGE was this advantage significant. The variance among the LS constants for SCA with barrows (table 3) did not approach significance for any trait. However, the SCA of inter-breed, intra-line crosses was highest for W56, W14 and ADG, and lowest for AGE, in accordance with results for gilts (table 2). Only for W21 in inter-breed, intra-line crosses (table 4) was any gilt - barrow difference in SCA significant. Analyses with odel 2 Pure Lines. ncluded in table 5 are LS comparisons among pure lines and among line crosses from separate analyses of the gilt and barrow data with model 2. Results from similar analyses of the gilt -- barrow data are presented in table 6. Among pure-line gilts, Yorkshires were slightly heavier (P.O) at W21, but Durocswere heavier at W14, younger in AGE and had more ABF and larger ADG than Yorkshires (P.5 or P.1). Among barrows, Durocs were younger (P.5) in AGE and had more (P.1) ABF than Yorkshires. With gilt - barrow data (table 6), Yorkshires exceeded Durocs in AGE by 5.84 d more (P.5) among gilts than among barrows; Durocs exceeded Yorkshires in ABF by.17 cm more (P.5) among gilts than among barrows; and Durocs exceeded Yorkshires in ADG by.3 kg/d more (P.5) among gilts than among barrows. n comparisons between high-fat and low-fat lines, gilts of the low-fat lines were heavier at birth, at W21 and W56, were older in AGE, had less ABF and smaller ADG than gilts in the high-fat lines (P. 1 to P.1). Among barrows, low-fat lines were heavier in all four weights, younger in AGE, had less ABF and higher ADG than barrows in the two high-fat lines. From gilt-barrow analyses, interactions of sex of pig with high- and low-fat lines were noted for W14, AGE and ADG (P.1). nteractions of breed line were noted for W14, AGE, ABF and ADG (P.5 or P.O1) with gilts, whereas with barrows, such interactions were noted only for W56 (P.5) and ABF (P.O1). With gilt - barrow data, interactions of sex of pig with breed x line effects were noted only for AGE and ADG (P.5). ntra-breed Line Crosses. Duroc intra-breed line crosses among both gilts and barrows exceeded those in Yorkshires (table 5) for W14, ADG and ABF, and were younger in AGE (P.5 or P.1). With the gilt - barrow data (table 6), intra-duroc crosses exceeded intra-yorkshire crosses for birth weight by.86 kg more (P.5) in gilts than in barrows, whereas intra-yorkshire crosses exceeded intra- Duroc crosses in AGE by 4.28 d more in gilts than in barrows (P.O).

9 DALLEL CROSSES OF HGH- AND LOW-FAT SWNE 43 Duroc intra-breed line crosses with DL dams were heavier than line crosses with DH dams for birth weight, W21 and W56 and had more ABF (P.O to P.1), among gilts. Among barrows, line crosses with DL dams were heavier than those with DH dams for W21, W56 and W14 (P.1 or P.5). Similar results were noted with intra-yorkshire line crosses, line crosses with YL dams exceeding those with YH dams. No significant interaction effects of sex of pigs with the reciprocal intra-breed line crosses were noted for any trait. nter-breed Line Crosses. With inter-breed, intra-line crosses, similar results were noted among gilts and barrows (table 5), with low-fat line crosses being heavier in all four weights, younger in AGE and with less ABF than high-fat line crosses (P.5 or P.O1), except that differences for W14 and AGE were nonsignificant among gilts. With the gilt - barrow data (table 6), low-fat line crosses exceeded high-fat line crosses in W14 by 2.99 kg more in barrows than in gilts (P.5). Also, for ABF, high-fat line crosses exceeded low-fat line crosses by.16 cm more (P.O) in gilts than in barrows. Between reciprocal crosses of the high-fat lines and of the low-fat lines, no significant differences were noted among gilts or barrows or in the gilt-barrow data for any trait except ADG in barrows, with the YH x DH cross exceeding the reciprocal cross by.27 kg/d (P.O). nter-breed, inter-line crosses with low-fat dams of either breed exceeded the reciprocal crosses with high-fat dams for the four weights and were younger in AGE (several significant, table 5) among both gilts and barrows. No significant gilt -- barrow differences were noted in comparisons between reciprocal low-fat x high-fat lines for any trait (table 6). Further, reciprocal crosses of DL YH exceeded reciprocal crosses of DH YL for the four weights and were younger in AGE among gilts (P.1 or P.5). Among barrows, reciprocal crosses of DH x YL exceeded those of DL x YH for W14 (P.1) and for ABF (P.5). Finally, only for ABF did the responses differ between gilts and barrows (table 6), with the reciprocal crosses of DH YL exceeding those of DL X YH by.2 cm more (P.5) in barrows than in gilts. GCA, aternal and Pure-Line Correlations. With model (1), GCA was estimated in part and maternal effects wholly from line of dam. Thus, while LS methods provided best linear unbiased estimates of effects in the assumed model, estimates of GCA and maternal effects were correlated. Correlations for each trait, computed from the respective LS constants for GCA (g) and maternal ability (m) of the four lines for gilts and barrows (tables 2 and 3) are shown as rg m in table 7. Correlations are shown for gilts and barrows separately and for the two sexes pooled. With so few df, only the high positive correlations involving pig weight at 21 d were statistically significant. Bereskin et al. (1974) also reported sizeable positive rg m values for average pig weight in litters at birth and at 21 d of age and rg m of -.33 for average pig weight in litters at 56 d, which is comparable to the pooled value of -.3 here for individual pig weight. Robison (1972) presented evidence for an apparent negative rg m in pig data, especially for preweaning litter traits. Jungst and Kuhlers (1984) reported significant negative values corresponding to rg m for litter size, total litter weight and litter average pig weight at birth and at 21 d of age. However, the present results (table 7) provide little, if any, consistent evidence for negative rg m values for individual growth rate in pigs. f the assumed model (1) is biologically correct, differences among pure lines (Plj) should provide unbiased estimates of line differences in GCA (g2j) and in maternal ability (m2j). The correlations (rpc, table 7) between the pure-line effect, Plj, and estimates of the corresponding effects (g2j + m2j) from line crosses were above.7 for all traits among gilts and barrows and in the pooled data. Despite the very few df, roe was statistically significant for all traits, with the pooled data. These results suggest that the average performance of a line among pure lines can provide a good indication of the line's average performance in line crosses. However, estimates of performance in a specific line cross would also require consideration of SCA for that cross and trait. Discussion Despite the obvious benefits of the diallel experimental design, few such experiments have been conducted with swine. These include analyses by Henderson (1948), Hetzer et al. (1961), Johnson et al. (1973), Bereskin et al. (1974),. Schneider et al. (1982) and the present

10 44 BERESKN AND HETER,.... ~ ~, m g m :.: :.: ~ 1 1 z e~ ~ ~ ~ ~ ~... ~7 ~" ~7 7 ~ 7~ z.~. ~..~. ~..~..~ : : " " " [- : ~. : [.-,.~xx~ ~Xx~ xx.~ ~ ~ ~ ~ ~.~ ~' ~=.oxx.'=xx..x "'~ ~'9' ~ "9' ~ x ~ ~ ~ ~'~

11 DALLEL CROSSES OF HGH- AND LOW-FAT SWNE 45 '1"1'.o ~. o.o. ~. o. ~.,.~ ~.~ "~ " " " e ~ ~ ~1'" ~ 9. ~ ~ TT ~ ~ T ~ r 9 ",,,-4 ~ " " '~ " e '']" X X X. ~ + ~ ~ ~ ~. ~,..u~ =~ 215 ~=,*, X x ~,, xx d ~. ~ x X x R ~.~ ~ ~ x xx d o

12 46 BERESKN AND HETER 9.. o.o.o, qo.o. o.o.o o.,~ ~.~o o~.o..t.-.v,~,r-i,r-i " i" ' " g 9 ~ U z z a.1 z o "1"1" o U t",l [',,,. t~ ~ Q r,,-~ t~ t~ O,.- oo.o.,.-, o,n o. ~. o. ~ o i " " " " " " U m, o. o.o. ooo. o.o.o o.o.o f E 8 ~,'r, ~ x x ~ X x ~ xx,.-, ',-.' "=,,:= ~ ~xx ~ xx.' ~xx ~V ~

13 DALLEL CROSSES OF HGH- AND LOW-FAT SWNE 47 analysis. Gregory et.al. (1978) and Stewart et al. (198) analyzed diallel experiments with beef cattle and Eisen et al. (1983) did so with mice. The present design was unique in that it included two breeds, each with two lines previously subjected to as many as 15 yr of divergent selection for either thicker or thinner backfat. n addition, inbreeding in the pure line pigs reached levels of 25 to 3%. The net effect was to elevate average heterosis effects in line crosses to significant levels for all weights and growth traits, but not for ABF. These results support the conclusion that heterosis is evidenced only in the presence of directional dominance effects at the loci affecting the trait and in differences in gene frequency between the lines being crossed (Falconer, 196). The divergence of gene frequencies can be assumed to be the joint result of inbreeding and selection practices. Thus, growth traits have been shown to be depressed by inbreeding and enhanced by crossing, while body composition traits such as backfat thickness have shown a nonsignificant response to inbreeding (Krehbiel et al., 1971) and little if any heterosis from crossing. Johnson (198) reported average heterosis of 9.4 and 2.5% for ADG and carcass ABF, respectively. Another useful finding here was the general lack of significant correlations between GCA and maternal effects for growth traits and ABF. This result contrasts with the largely negative correlations of these two parameters for pre- weaning litter traits (Bereskin et al., 1974; Falconer, 1965; Robison, 1972). Also of note were the high, significant correlations of pure-line effects with crossqine joint GCA and maternal effects, indicating that average performance of a line in crosses could largely be predicted from its performance as a pure line. n addition, SCA was generally not a significant factor in line crosses, except for pig weight at 21 d of age, and then only marginally. n intra-breed and in inter-breed, inter-line comparisons between reciprocal crosses (tables 5 and 6), pigs with low-fat dams performed better than pigs with high-fat dams for the four weights and age at 79.4 kg. These results suggest that the long-term selection for either thinner or thicker backfat apparently resulted in significant changes in maternal abilities of the different lines and in differences in correlated responses of growth rate to the selection practiced, in general accord with findings by Hetzer and iller (1972). Also, differences in sex effects (gilts vs barrows) were noted mainly among pure lines and then only in traits measured postweaning. Only isolated differences in sex effeets were noted among line crosses or with maternal effects, GCA or SCA. n summary, with the increasing interest in breed and line crossing in swine, it would appear prudent to utilize the diauel design in evaluating different breeds and lines for the various genetic para- meters that performance. affect crossbred or line-cross TABLE 7. CORRELATONS NVOLVNG PURE LNES (rpc) AND ATERNAL AND GENERAL COBNNG EFFECTS ~rgm) rgm Pig trait Gilts Barrows Pooled a Gilts Barrows Pooled a rpc Birth wt *.891 * * 21-d wt *.849"* ".895 * * 56-d wt * 14-d wt * * Age(d) at probe " Avg backfat thickness * 98 *.974 * * Avg daily gain * Error df apooled df was based on the eight comparisons overall rather than pooled within each sex. *P.5. **P.1.

14 48 BERESKN AND HETER Literature Cited Andrews, F. N. and R.. Whaley easure of fat and muscle in live animal and carcass. ndiana Agr. Exp. Sta. Annu. Rep. 68: Bereskin, B., H. O. Hetzer, W. H. Peters and H. W. Norton Genetic and maternal effects on pre-weaning traits in crosses of high- and low-fat lines of swine. J. Anim. Sci. 39:1. Bereskin, B. and H. W. Norton An adjustable algorithm for the analysis of data by least-squares. (ongr.) USDA, BeltsviUe, D. Dickerson, G. E nbreeding and heterosis in animals. n: Proc. of Animal Breeding and Genetics Symp. in Honor of Dr. J. L. Lush. Amer. Soc. Anita. Sci., Champaign, 1L. Eisen, E. J., G. Horstgen-Schwark, A.. Saxton and T. R. Bandy Genetic interpretation and analysis of diallel crosses in animals. Theor. Appl. Genet. 65:17. Falconer, D. S ntroduction to quantitative Genetics. Ronald Press Co., New York, NY. Falconer, D. S aternal effects and selection response. n: S. J. Geerts (Ed.) Genetics Today. p 26. Pergamon Press, New York, NY. Gregory, K. E., L. V. Cundiff, R.. Koch, D. B. Laster and G.. Smith Heterosis and breed maternal and transmitted effects in beef cattle.. Preweaningtraits. J. Anim. Sci. 47:131. Harvey, W. R Least-squares analysis of data with unequal subclass numbers. USDA, ARS H-4. Henderson, C. R Estimation of general, specific and maternal combining ability in crosses among inbred lines of swine. Ph.D. Thesis. owa State Univ., Ames. - Henderson, C. R Specific and general combining ability. n: John B. Gowen (Ed.) Heterosis. Haefner Publ. Co., New York, NY. Hetzer, H. O., R. E. Comstock, J. H. eller, R. L. Hiner and W. R. Harvey Combining abilities in crosses among six inbred lines of swine. USDA, Tech. Bull Hetzer, H. O. and W. R. Harvey Selection for high and low fatness in swine. J. Anim. Sci. 26:1244. Hetzer, H. O. and R. H. iller Rate of growth as influenced by selection for high and low fatness in swine. J. Anim. Sci. 35:73. Johnson, R. K Heterosis and breed effects in swine. North Central Regional Pub Agr. Exp. Sta., Univ. of Nebraska, Lincoln. Johnson, R. K.,. T. Omtvedt and L. E. Wahers Evaluation of purebreds and two-breed crosses in swine: Feedlot performance and carcass merit. J. Anim. Sci. 37:18, Jungst, S. B. and D. L. Kuhlers Estimates of additive genetic, maternal and specific combining abilities for some litter traits in swine. J. Anita. Sci. 59:114. Krehbiel, E. V., H. O. Hetzer, A. E. Flower, G. E. Dickerson, W. R. Harvey and L. A. Swiger Effectiveness of reciprocal selection for performance of crosses between ontana No. 1 and Yorkshire swine. f. Postweaning traits. J. Anim. Sci. 32:211. Robison, O. W aternal effects in swine. J. Anim. Sci. 35:133. Schneider, J. F., L. L. Christian and D. L. Kuhlers Crossbreeding in swine: Genetic effects on pig growth and carcass merit. J. Anim. Sci. 54:747. Stewart, T. S., C. R. Long and T. C. Cartwright Characterization of cattle of a five-breed diallel. ll. Puberty in bulls and heifers. J. Anita. Sci. 5:88.