Discussion
The F line apparently had a different growth curve than the two commercial sire lines. At 8 weeks of age, the F line had similar body weight or was heavier than the sire lines but was smaller than both commercial lines at older ages. The two commercial sire lines may have also differed in their growth curves. The difference in body weight between the F and A lines increased to a greater extent from 16 to 20 weeks of age than the difference between the F and B lines. Based on commercial birds of the same breeders from which the sire lines were obtained in the present study, Barbour and Lilburn (1996) observed a difference in growth pattern of males through 82 days of age. However, in an earlier study using commercial females of the same breeders, Barbour and Lilburn (1995) did not observe an interaction of the two commercial strains and age for body weight from 14 to 145 days, suggesting no difference in growth pattern between the commercial lines. In comparisons of commercial birds from two major turkey breeders, one of which was not involved in the current study, differences in growth curves were noted for both males (Lilburn and Emmerson, 1993) and females (Lilburn et al., 1992).
Additive genetic variation, as indicated by differences among lines, was an important source of variation, as expected, for body weight at 16 and 20 weeks of age, shank length, shank width (females only), and breast width. It has been shown that large gains in body weight (McCartney et al., 1968; Nestor, 1977b, 1984; Nestor et al., 1996), breast width (Nestor et al., 1969), and shank width (Nestor et al., 1985) can be made by selection.
Nonadditive genetic variation was evident for body weight of males at 8, 16, and 20 weeks of age in the crosses of the F line and two commercial sire lines. In females, such variation was important only at 8 or 16 weeks of age. Age specific heterosis in body weight has been observed previously (Asmundson, 1942; Asmundson and Pun, 1954). Significant heterosis was also observed in shank length, shank width, and shank depth in some comparisons, and walking ability score for males from the crosses of the F and B lines.
The F line is not closely related to the commercial sire Lines A and B even though the F line was started from a randombred control that was developed from reciprocal crosses of two commercial strains of turkeys (Nestor et al., 1969). The F line differed greatly from Lines A and B in the frequency of major histocompatibility complex Class II haplotypes (Zhu et al., 1995, 1996b) and had a band sharing of DNA fingerprints with Lines A or B that was less than average of all lines compared (Ye et al., 1998). Band sharing of DNA fingerprints of Lines A and B was greater than the average of all lines in the study of Ye et al. (1998). Inbreeding, as measured by band sharing of DNA fingerprints (Kuhnlein et al., 1990; Zhu et al., 1996a), was greater in the commercial sire lines than in the F line (Ye et al., 1998). Accumulated inbreeding in the F line, as estimated by variation in family size, was 27.6% (unpublished data) when the crosses were made with the commercial sire lines. Therefore, because the lines involved in the crosses were moderately inbred, the heterosis observed in the present study may have been due to the elimination of inbreeding effects by crossing relatively unrelated lines. Based on a simple dominance model, a linear relationship is expected between the degree of heterosis and the level of inbreeding (Hill, 1982). The magnitude of heterosis is inversely related to the degree of genetic resemblance between parental populations (Wilham and Pollak, 1985).
Nonadditive genetic variation in body weight of turkeys is usually not an important source of genetic variation (Kondra and Shoffner, 1955; Clark, 1961; McCartney and Chamberlin, 1961; Nestor, 1971), but heterosis has been observed in some crosses in which the parents differed greatly in body conformation (Asmundson, 1945, 1948; Emmerson et al, 1991; Ye et al., 1997; Nestor and Anderson, 1998). There were large differences in breast width between the F and A and F and B lines and the lines differed in body shape (unpublished data). The differences in conformation and body shape between the experimental F line and the commercial sire lines may have been responsible for the heterosis in body weight observed. It is unknown why heterosis was expressed to a greater extent in males than in females.
No heterosis in breast width was observed in the current study. Similarly, Asmundson (1948), Ye et al. (1997), and Nestor and Anderson (1998) did not observe heterosis in crosses of turkey lines differing greatly in breast width. Using diallel crosses among six strains, McCartney and Chamberlin (1961) found that the nonadditive genetic variation accounted for 8 % of the total genetic variation in breast width of males but there was no nonadditive genetic variation in breast width in females.
In summary, nonadditive genetic variation was a major source of variation in body weight of males from reciprocal crosses of an experimental growth line and two commercial sire lines. The amount of heterosis ranged from 3.1 to 7.5% at 8, 16, and 20 weeks of age. For females, heterosis was present in the crosses only at 8 or 16 weeks of age. Significant heterosis was observed for shank length and, in some cases for shank width and shank depth. No significant heterosis was observed for breast width.