- Email: [email protected]
Estimates of Genetic Parameters in Turkeys.
Estimates of Genetic Parameters in Turkeys. 2. Body Weight and Carcass Characteristics 1 G. B. HAVENSTEIN, V. D. TOELLE, K. E. NESTOR, and W. L. BACON
(Received for publication January 19, 1988) ABSTRACT Data were taken from 1,088 individuals-504 females (F) and 584 males (M)-produced by 34 sires and 168 dams from a randombred control line (RBC2) of turkeys maintained at the Ohio Agricultural Research and Development Center, Wooster, OH. Turkeys were used to estimate genetic and phenotypic parameters among 16-wk body weight and a number of body composition characteristics. Heritabilities (h2) were estimated from sire components of variance for 16-wk weights of muscles of breast (BM: F = .08, M = .35); thigh (TM: F = .12, M = .17); and drum (DM: F = .30, M = .44); weights of leaf fat (LF: F = .13, M = .51) and total abdominal fat (AF: F = .21, M = .55); and total percentage carcass fat (%F: F = .06, M = .24), percentage protein (F = .07, M = .20), and percent moisture (F = nonestimable, M = .11). The h2 of percentage carcass ash was nonestimable from both data sets due to negative estimates of the sire variance components. Due to the non-normal distributions of some of the fat measures, the fat measure data were converted to natural logarithms. Genetic estimates from the transformed data were similar to those from the nontransformed data. Genetic correlation (rG) estimates indicated a positive genetic association between body weight (BW) and all the fat characteristics measured. The BW was highly correlated with BM (F = .48, M = .86), TM (F = .81, M = 1.01), and DM (F = 1.04, M = .70). However, BW was correlated to a lesser degree with LF (F = .02, M = .27), AF (F = .25, M = .32), and %F (F = .93, M = .50). The BM was only moderately correlated with TM (F = .43, M = .28) and DM (F = .73, M = .27). The TM and DM were highly correlated (F = 1.02, M = .84). The existence of these correlations, which are less than unity, indicate that selection for BW alone will lead to increasing levels of fat in commercial turkeys, and to disproportionate increases in breast vs. leg muscles. (Key words: turkeys, body weight, heritability, genetic correlation, carcass composition) 1988 Poultry Science 67:1388-1399 INTRODUCTION
The goal of any food production enterprise is to produce consumer acceptable products of high quality at the most competitive price. Over the years, breeders of meat type poultry have concentrated their genetic improvement efforts on selection for heavier body weights and wider breast conformation at a given market weight. Due to the highly heritable nature of growth rate, their efforts have resulted in vastly improved growth rates for turkey and broiler populations. The effectiveness of their efforts on the growth rate of turkeys has been well documented in a review by Arthur and Abplanalp (1975), and in recent reports by Nestor (1984) and by Delabrosse et al. (1986).
'Salaries and research support provided by state and federal funds appropriated to the Ohio Agricultural Research and Development Center, The Ohio State University. Journal Article Number 3-88.
Traditionally, changes brought about in growth rates of chickens and turkeys have been accomplished without paying much attention to body composition. It is now well documented that increased growth rate has been accompanied by undesirable increases in body fat (for reviews, see Lin et al., 1980; Leenstra, 1986). Numerous workers have also shown that the genetic correlation between body weight and fat deposition, at least in broilers, is positive (Ricard and Rouvier, 1967, 1969; Proudman et al., 1970; Wethli and Wessels, 1973; Becker, 1978; Becker et al., 1984) and that selection towards increased growth rate has resulted in excessive amounts of fat at slaughter age in commercial broilers (Nordstrom et al., 1978; Marks, 1979; Chambers et al., 1981; Becker et al., 1984; Siegel, 1984; Leenstra et al., 1986; Leenstra and Pit, 1987). Some of these studies have also shown that different genetic stocks differ in the amount of fat present at market weight.
1388
Downloaded from http://ps.oxfordjournals.org/ at Ernst Mayr Library of the Museum Comp Zoology, Harvard University on April 20, 2015
Department of Poultry Science, The Ohio State University, Ohio Agricultural Research and Development Center, Wooster, Ohio 44691
GENETIC COMPOSITION PARAMETERS IN TURKEYS
MATERIALS AND METHODS
Data were collected at 16 wk of age on 1,088 fully pedigreed turkeys (504 females and 584 males) of a randombred control line (RBC2) maintained at the Ohio Agricultural Research and Development Center, Wooster, OH. The data reported herein are part of a larger study that also included measurements on a number of skeletal characteristics. The analyses of those data were reported in the first paper in this series (Havenstein et al., 1988), which also included a complete description of the materials and methods. As reported in the companion paper, one-half of the carcass was used for whole body analyses
of percentage carcass fat (%F), percentage ash (%A), percentage protein (%P), and percentage moisture (%M). The other half of the carcass was cut up and the breast muscles (BM), thigh muscles (TM), and drum muscles (DM) were weighed. Leaf fat (LF) and total abdominal fat (AF) were also weighed and recorded. Before analysis, muscle weights were multiplied by two, and they and AF and LF were used to calculate percentages of total live weight (%BM, %TM, %DM, %LF, and %AF). Thus, analyses of fat and of muscles were made on both a weight and a percentage of live weight basis. Sixteen-week body weight (BW), shank width (SW), and shank length (SL) were reported in the companion paper (Havenstein et al., 1988), but are also included herein as a common reference point. Because a preliminary analysis indicated that AF and LF measurements were not normally distributed, those measurements were transformed to natural logarithms before analysis. The transformed traits are referred to as LNAF and LNLF, respectively. The %F was also transformed in this manner and is reported as LN%F. Data were analyzed and variance and covariance components were estimated using the least-squares and maximum likelihood procedures outlined by Harvey (1986). The data set containing both sexes was analyzed using the following model: Yijklm = JJL + Si + d/Sij + Hk + SX, + bCXjjkim - x) + eijklm where: |x = the overall mean for trait Y; Sj = the random effect of the ith sire; d/Sy = the random effect of the j t h dam mated to the ith sire; Hk = the fixed effect of the kth hatch; SX, = the fixed effect of the 1th sex; b = the regression of Yyklm on the day of kill; Xyklm = the day of kill for the individual; x = the average day of kill; and eijkim = the random error associated with the measurement of each individual, which is assumed to be randomly and independently distributed with a mean of zero and a variance of a2. The data were also analyzed separately by sex using the same analysis as shown above except that sex was deleted from the model. This was done because turkeys have an extreme degree of sexual dimorphism for body, muscle, and fat weights, which results in large differences between means and variances for the two
Downloaded from http://ps.oxfordjournals.org/ at Ernst Mayr Library of the Museum Comp Zoology, Harvard University on April 20, 2015
Several investigators (Johnson and Gowe, 1962; Nestor, 1982; Bacon et al, 1986) have demonstrated that the amount of body fat has also increased with growth rate selection in turkeys, but as yet it is not considered to be a significant problem. However, there is increasing concern that current selection practices by the turkey breeding industry will eventually lead to the same consumer complaints about excessive fatness that the broiler industry has had to face over the past few years. Excessive fat levels in meat type poultry is a major concern. It leads both to economic losses for processors, who send much of the abdominal fat along with the viscera to the rendering plant, and to the consumer, who generally discards it. Large amounts of fat left in the carcass also lead to complaints and decreased consumer acceptance. It has been postulated (Nestor et al., 1985) that selection for increased growth rate has led to disproportionate increases in the relative amounts of breast vs. leg muscles, and that this disproportionate change has brought about at least part of the leg weakness problem that the turkey industry is currently facing. Only one study is available at this time on the genetic relationships among body weight, various body parts, and fat deposition in turkeys (Delabrosse et al., 1986). The purpose of the present study was to examine the inheritance of and the genetic and phenotypic relationships among live body weight and a number of muscle and fat measurements in turkeys, as well as genetic and phenotypic relationships of those traits with shank width and length. The latter traits appear to be a useful measure for the improvement of the skeletal support system (Nestor et al., 1987; Havenstein et al., 1988).
1389
HAVENSTEIN ET AL.
1390
RESULTS AND DISCUSSION
Testsfor Heterogeneity of the Variancesfrom the Within Sex Analyses. Phenotypic variances for the same' 'trait'' from the within sex analyses were found to be heterogeneous for all of the traits except %A, %M, %BM, %TM, %DM, and LN%F. Thus, the data from males (M) and females (F) for the majority of traits should be considered separately. However, because the ANOVA is a very robust technique, genetic parameter estimates were also made from the com-
bined-data set and are provided herein for comparison purposes. Means and Phenotypic Standard Deviations. The overall means and the phenotypic standard deviations from both within-sex and combineddata set analyses for BW, S W, SL, all the muscle and fat weights, all the muscle and fat percentages, and the transformation of fat measures to natural logarithms are presented in Table 1. The M had significantly higher (P<.05) mean values for BW, SW, SL, BM, TM, DM, %P, %A, and %M than F. However, F had significantly higher (P<.05) mean values for all fat measures and muscle traits when expressed as a percentage of body weight. These observations are consistent with results of most carcass composition studies, where M generally have larger absolute values for body and muscle weights, but F have larger values for fat measures and percentage muscle (Harshaw and Rector, 1940; Fry et al., 1962; Hartung and Froning, 1968; Leeson and Summers, 1980). Heritability Estimates. The h 2 estimates for BW, SW, SL, and the carcass characteristics
TABLE 1. Overall means and standard deviations (op) for traits studied as calculated within sex and with sexes combined Females
Males Trait 1 BW, kg SW, mm SL, cm
op
X
7.34
.56
Sexes combined op
X
5.25
op
X
.42
6.37
.50
13.3 15.6
.5 .5
11.8 12.8
.5 .5
12.6 14.3
.5 .5
665.0 347.2 260.7
73.2 41.2 29.8
489.0 262.9 200.3
56.8 32.6 22.8
583.5 308.1 232.7
66.1 37.5 26.8
3.0 6.5
3.1 5.4
7.9
7.2
5.2
15.4
12.7
10.6
5.4 9.6
%F %A %P %M
13.0 12.6 68.0 70.8
3.0 2.5 3.5 1.8
17.9 11.8 63.9 69.0
4.2 2.5 4.3 1.8
15.3 12.2 66.1 70.0
3.6 2.5 3.9 1.8
%BM %TM %DM
18.1
1.2 .9 .6
18.6 10.0
1.5 .9 .6
18.3
1.4 .9 .6
BM, g TM, g DM, g LF,g AF,g
%LF %AF
LNLF LNAF LN%F
9.5 7.1
7.6
9.7 7.3
.04 .09
.04 .07
.15 .29
.13 .23
.09 .18
.09 .16
1.17 1.84 2.61
.59 .56 .22
1.89 2.51 2.91
.76 .76 .22
1.50 2.15 2.75
.68 .66 .22
1 SW = Shank width; SL = shank length; BM, DM, and TM = breast, drum, and thigh muscles, respectively; LF and AF = leaf fat and total abdominal fat, respectively; %F, %A, %P, and %M = % fat, % ash, % protein, and % moisture, respectively; %BM, %TM, %DM, %LF, and %AF = muscle and fat traits expressed as a % of live BW, respectively; LNLF, LNAF, and LN%F = natural logarithms of LF, AF, and %F, respectively.
Downloaded from http://ps.oxfordjournals.org/ at Ernst Mayr Library of the Museum Comp Zoology, Harvard University on April 20, 2015
sexes. Phenotypic variances for the same "trait" from these within sex analyses were tested for heterogeneity using the method of Bartlett (1937) as shown by Neter and Wasserman (1974). The results are reported from both the within-sex and the pooled-data set analyses. Standard errors for the heritability (h2) and genetic correlation (rG) estimates were calculated by modifications of the methods of Tallis (1959) and Swiger et al. (1964) as described by Harvey (1986).
GENETIC COMPOSITION PARAMETERS IN TURKEYS
culated separately for the two sexes, but differences between their estimates tended to be in the opposite direction to those reported herein. They, however, slaughtered M and F at different ages, whereas M and F were slaughtered at the same ages in this study. This difference in procedure, and in the estimates obtained, indicates that differences in physiological development at a common killing age may affect the expression of the "same trait" differently in the two sexes. If so, then a "trait" measured at the same chronological age in the two sexes may actually be two different traits, and therefore, should be included separately in genetic selection programs so as to maximize the improvement of overall genetic merit. The ex2 h2 estimates for BM, TM, and DM were higher than those for their comparable percentages of BW values, i.e., %BM, %TM, and %DM. Fat characteristics transformed to natural logarithms also had higher h 2 estimates from the
TABLE 2. Heritability estimates (h2) and their standard errors (SE) from sire (CT|) and dam (afi) components of variance when measured within sex and with sexes combined Sexes c o m b i n e d
Females
Males
°s
°s
°y
°d
"s
fd
Traits 1
h2
SE
h2
SE
h2
SE
h2
SE
h2
SE
h2
BW
.60
.19
.63
17
.23
.14
1.22
.20
.45
.13
.82
.12
SW SL
.47 .54
.17 .19
.48 .58
17 17
.55 .43
.20 .18
.68 .06
.20 .18
.46 .51
.14 .14
.53 .27
.11 .10
BM TM DM
.35 .17 .44
.15 .11 .17
.87 .77 .39
18 18 17
.08 .12 .30
.10 .11 .15
.91 .80 .84
.20 .20 .20
.25 .21 .35
.09 .08 .11
.86 .64 .49
.13 .12 .11
LF AF
.51 .55
.18 .19
.57 .61
17 17
.13 .21
.11 .13
.78 .92
.20 .20
.21 .27
.08 .10
.55 .58
.11 .11
%F %A %P %M
.24 NE2 .20 .11
.12
17 16 16 15
.06 NE .07 NE
.10
.68 .18 .61 .80
.20 .18 .20 .20
.16 NE .16 .08
.07
.12 .09
.60 .19 .31 .08
.07 .05
.46 .05 .21 .13
.11 .08 .09 .09
%BM %TM %DM
.23 .00 .36
.12 .07 .15
.72 .68 .38
18 18 17
.19 .07 .04
.13 .10 .09
.20 .48 .50
.19 .19 .19
.19 .11 .22
.08 .06 .09
.44 .22 .27
.11 .09 .10
%LF %AF
.51 .56
.18 .19
.59 .64
17 17
.14 .21
.12 .13
.71 .87
.20 .20
.18 .23
.08 .09
.48 .51
.11 .11
LNLF LNAF
.64 .63
.20 .20
.41 .48
17 17
.20 .28
.13 .15
.68 .81
.20 .20
.38 .41
.12 .13
.55 .60
.11 .11
LN%F
.25
.13
.56
17
.04
.09
.70
.20
.18
.08
.45
.11
.10
1
SE
SW = Shank width; SL = shank length; BM, DM, and TM = breast, drum, and thigh muscles, respectively; LF and AF = leaf fat and total abdominal fat, respectively; %F, %A, %P, and %M = % fat, % ash, % protein, and % moisture, respectively; %BM, %TM, %DM, %LF, and %AF = muscle and fat traits expressed as a % of live BW, respectively; LNLF, LNAF, and LN%F = natural logarithms of LF, AF, and %F, respectively. 2
NE = Nonestimable, due to a negative variance component estimate.
Downloaded from http://ps.oxfordjournals.org/ at Ernst Mayr Library of the Museum Comp Zoology, Harvard University on April 20, 2015
are summarized in Table 2. The BW, SW, and SL estimates were reported in the companion paper (Havenstein et al., 1988), but are also reported herein because of their overall importance to turkey breeding programs, and to show their relationships with the traits in this portion of the study. The h 2 estimates for the muscle and fat characteristics show a similar pattern to those for skeletal characteristics reported by Havenstein et al. (1988), in that the estimates from sire components of variance (a 2 ) from M data are generally higher than those from F data. However, h estimates from dam components of variance (a2,) were higher in F than in M data. These results support the conclusion by Havenstein et al. (1988) that the expression of genetic variation for the same trait at the same killing age is influenced by the sex of the individuals involved. However, Delabrosse et al. (1986) also reported h 2 estimates that were cal-
1391
1392
HAVENSTEIN ET AL.
ficients of variation for fat measures in the range of 25 to 30%, whereas the turkey data reported herein show coefficients of variation for LF and AF approaching 100% (see the means and standard errors in Table 1). The question is, which of these measures of fatness provides the best criterion for selection purposes? It appears from this study that %F provides the best information from a statistical point of view, because it is normally distributed, and it also has the highest correlation with BW. However, from an application point of view, as measurement of %F requires the bird to be sacrificed and a complete carcass analysis performed, it would be difficult and expensive to use %F as a routine selection criterion for a commercial turkey breeding program. It also appears that in spite of the non-normality of the distributions for LF and AF, that LF, AF, and %F were highly correlated genetically (r G > .95), and that they are, therefore, equally good measures of fatness. However, because all of the measures used in this study require the use of sacrificed sibs to obtain the selection data, it appears that until a better measure is found, selection for reduced levels of the very low density lipoproteins as suggested by Griffin and Whitehead (1985) is the method of choice for reducing fatness in turkeys. Griffin and Whitehead's method may be less costly to use, and probably provides a better criterion for selection than any of the measures reported in this study. It is certain, however, based on the rG present in these data and in the data of Delabrosse et al. (1986), that continued selection for BW and breast conformation score in turkeys, a measure of total breast muscle, will eventually result in excessively fat turkeys, just as it has with broiler chickens. The other whole body analyses, i.e., for %P, %A, and %M, also showed rG with BW (Tables 3 , 4 , and 6). Negative of estimates from both the M and F data prevented the actual calculation of rG estimates for %A. The r P for these two traits was -.11 for both sexes. All estimates of the rG between BW and %P and %M were negative, as were the rP between these traits. Again, a negative of from the F data prevented the calculation of an estimate for %M from that data set. Nevertheless, it appears that as BW is increased by selection, both %P and %M will decrease. Estimates of the rG between SW and BW were positive and relatively low (Havenstein et al, 1988). The rG between SW and BM, TM,
Downloaded from http://ps.oxfordjournals.org/ at Ernst Mayr Library of the Museum Comp Zoology, Harvard University on April 20, 2015
of than did their comparable nontransformed measurements. These h patterns were observed in estimates from both sexes (Table 2). Estimates of Genetic (rG) andPhenotypic (rP) Correlations. Summaries of the rG and r P correlations, calculated from the sire components of variance and covariance from the within-sex analyses, can be found in Tables 3 and 4. Standard errors of rG estimates in Tables 3 and 4 are presented in Table 5. The rG and rP estimates from the combined-data set analysis are provided in Table 6, and standard errors for the rG estimates from the combined analysis are given in Table 7. The rG estimates of BW with BM, TM, and DM weights tended to be high (range .48 to 1.04) in both sexes (Tables 3 and 4). The TM and DM weights of F appeared to be somewhat more highly correlated with BW than the BM weights. Estimates from these same muscle weight groups, when expressed as a percentage of live weight, are far less clear cut. For example, the rG of BW with %BM and %TM were negative in both sexes, whereas correlations with %DM were negative from the M data and highly positive from the F data. Extreme caution should be used in evaluating and using correlation estimates that are based on a ratio trait, especially one which is being correlated with a trait that is one of its component parts. Gunsett (1984) and Brown et al. (1985) discussed some of the inherent problems associated with using ratio or percentage traits in which the numerator and denominator are moderately to highly correlated. Genetic correlation estimates of B W with all fat measures were positive in both sexes (Tables 3 and 4). The LF showed the lowest rG with BW (F = .02, M = .27). Total AF was slightly (F = .25, M = .32), and %F was considerably more highly correlated (F = .93, M = .50) with BW than was LF. Fat measures were also expressed as a percentage of body weight, and as natural logarithm transformations of the original measures. The rG of BW with the transformed fat measures were very similar to those from their comparable nontransformed measures (Tables 3 , 4 , and 6). Transformation to natural logarithms was done because LF and AF are not normally distributed and are extremely skewed in the upward direction. Becker et al. (1984) reported a similar situation in broilers, but the non-normality and skewness is even greater in the present turkey data than in the broiler data. For example, Becker et al. (1984) reported coef-
GENETIC COMPOSITION PARAMETERS IN TURKEYS
o o CO 00 ^
^+
1
rt
" C\l
o
z
1 Z
Z
w 1 z
-
1
o in o
in
00
o
NO W
fN} o 00 fN] 1 1
1
ON O 00
o
o *t o
o o
1
1
r~
00 NO 00 ON O N m
t
~
CO NO NO CO CO ON O N
o o
00 ON 1 ^ O ON O
1 z
Z
1
1
rt^
1
ON
w Z
~
ON
1 1
Z
O l " ~ f ^ - O © O 0 \ O < ' - I f • rt • rA W I I 1 Z i *o t-> H r n CO 0 y M 00 • n rt O : i i i Z 1
O O O N O f i N C O N N "* f f rt U <-i U] 1 I Z Z r^so^O"%'Hr^GOoo^ f
f
a*
l ' I H M M U r i [I] I I I Z Z
I
Z Z Z Z Z Z Z Z Z Z Z
I
I
I I
U U U t i U U K U U Z Z Z Z Z Z Z Z Z
I I I
I _
U, tin U.
S S j S S S t t . n . t < a . S e o H Q J < Z Z Z
Downloaded from http://ps.oxfordjournals.org/ at Ernst Mayr Library of the Museum Comp Zoology, Harvard University on April 20, 2015
00 (N C-l NO
NO O N CO
7
.35 -.60
-H
.29 .24 .24 -.01 .02 -.03 .10 -.01 -.10 .10 .03 .03 -.03 -.01 .00 .03 -.03
.53 .32
SL
.63 .44 .10 .13 .20 -.11 -.09 -.03 .73 .15 -.10 .05 .06 .12 .15 .19
.86 .41 .34
BM
.56 .14 .16 .15 -.07 -.08 -.07 .32 .76 .19 .10 .11 .14 .17 .15
1.01 .29 -.07 .28
TM
.17 .18 .16 -.05 -.08 -.03 .04 .22 .75 .13 .14 .18 .20 .16
.70 .30 .00 .27 .84
DM
.95 .57 -.12 -.45 -.25 -.09 -.01 .02 .99 .95 .93 .88 .54
.27 .21 .05 .16 .35 .40
LF
.62 -.14 -.47 -.26 -.10 -.02 .00 .94 .99 .88 .93 .58
.32 .21 .05 .19 .41 .39 1.00
AF
-.19 -.36 -.04 .01 .02 -.11 -.13 -.13 -.16 -.39 .42 .07 .06 .07 -.44 -.46 -.45 -.48 -.59
-.23 .10 .04 .06 -.26 -.49 -.92 -.98 -1.16 NE
NE3 NE NE NE NE NE NE NE NE
.50 .06 .34 .39 .62 .34 1.00 1.03 -.40 -.60 -.26 -.04 -.07 -.06 .55 .61 .59 .66 .99
%P
%A
%F
(rp) correlation estimates from sire components %BM
-.25 -.11 -.27 .18 -.55 -.30 .42 -.09 -.43 -1.30 -.58 -.76 -.31 -.16 -.35 -.17 -.47 -.09 NE NE .47 .08 .30 .06 -.01 .28 .04 -.09 -.25 -.10 -.25 -.11 -.26 -.09 -.26 -.11 -.26 -.05
%M
.30 -.0 -.02 -.0 -.04 -.08
-1.84 -1.54 -7.50 -7.96 -2.04 .14 .33 .40 .65 NE -.04 -1.48 -12.82
%TM
of variance and covar
3
N E = Nonestimable, due to negative variance component.
SW = Shank width; SL = shank length; BM, DM, and TM = breast, drum, and thigh muscles, respectively; LF and AF = leaf f %M = % fat, % ash, % protein, and % moisture, respectively; %BM, %TM, %DM, %LF, and %AF = muscle and fat traits expressed a natural logarithms of LF, AF, and %F, respectively.
3
.09 .28 .27 .40 .05 .06 .07 .03 -.06 -.11 .02 .02 .18 .03 .02 .03 .04 .07
.33
SW
The TQ are above and rp are below the diagonal.
.39 .32 .79 .62 .61 .23 .28 .32 -.11 -.20 -.09 .16 -.03 -.07 .17 .20 .25 .31 .32
BW SW SL BM TM DM LF AF %F %A %P %M %BM %TM %DM %LF %AF LNLF LNAF LN%F
1
BW
Trait 2
TABLE 4. Genetic (re) and phenotypic
djournals.org/ at Ernst Mayr Library of the Museum Comp Zoology, Harvard University on April 20, 2015
.23 .31 .25 .24 .24 .27 NE .30 .36 .27 NE .21 .24 .24 .23 .23 .26
.19 .26
SL
.30 .25 .26 .25 .27 NE .32 .39 .24 NE .25 .26 .26 .25 .24 .27
.37 .36 .41
BM
.15 .29 .28 .31 NE .39 .48 .22 NE .33 .30 .29 .29 .27 .30
.19 .23 .37 .49
TM
.22 .22 .27 NE .29 .36 .23 NE .15 .23 .23 .22 .21 .26
.08 .23 .25 .32 .15
DM
.01 .09 NE .30 .39 .29 NE .25 .00 .01 .01 .01 .09
.48 .38 .39 .67 .49 .42
LF
.07 NE .29 .38 .29 NE .25 .01 .00 .01 .01 .08
.36 .30 .30 .50 .45 .34 .04
AF
NE .44 .48 .34 NE .31 .09 .08 .08 .07 .00
.68 .78 .75 1.22 .79 .74 .43 .62
%F
.45 .33 NE .30 .30 .29 .28 .28 .43
1.07 1.04 .71 1.49 .90 .92 1.68 1.52 NE NE
NE3 NE NE NE NE NE NE NE NE NE NE NE NE NE NE NE NE NE NE
%P
%A
.43 NE .37 .39 .38 .37 .37 .47
NE NE NE NE NE NE NE NE NE NE NE
%M
NE .24 .29 .29 .28 .28 .34
.35 .29 .33 .46 .41 .33 .58 .41 1.86 NE 1.76 NE
%BM
NE NE NE NE NE NE
.52 .41 .47 .73 .41 .46 .67 .55 1.32 NE 1.76 NE .53
%TM
NE = Nonestimable, due to negative variance component.
''NP = Not presented because of standard errors > 2.00 due to sire components near zero.
3
SW = Shank width; SL = shank length; BM, DM, and TM = breast, drum, and thigh muscles, respectively; LF and AF = leaf %M = % fat, % ash, % protein, and % moisture, respectively; %BM, %TM, %DM, %LF, and %AF = muscle-and fat traits expressed a natural logarithms of LF, AF, and %F, respectively.
2
.23 .23 .29 .23 .24 .24 .29 NE .31 .38 .29 NE .26 .24 .24 .24 .24 .29
.24
SW
of variance and covariance for data on fem
Standard errors for the TQ estimates from the female data {Table 3) are above the diagonal, and for the TQ estimates from the
.21 .18 .07 .11 .13 .22 .21 .22 NE .30 .36 .28 NE .25 .23 .22 .21 .20 .21
BW SW SL BM TM DM LF AF %F %A %P %M %BM %TM %DM %LF %AF LNLF LNAF LN%F
1
BW
Trait 2
TABLE 5. Standard errors of genetic correlation (rQ) estimates from sire components
ournals.org/ at Ernst Mayr Library of the Museum Comp Zoology, Harvard University on April 20, 2015
.25 .22 .26 .02 .04 -.05 .07 .01 -.03 .03 .00 .06 -.01 .01 -.02 .01 -.06
.60 .18
SL
.62 .47 .10 .12 .15 -.11 -.05 -.05 .72 .17 -.05 .06 .07 .13 .15 .16
.82 .50 .44
BM
.59 .18 .19 .14 -.06 -.08 -.09 .26 .76 .21 .13 .15 .18 .21 .15
.85 .38 .07 .34
TM
.13 .14 .12 -.05 -.06 -.05 .06 .24 .73 .09 .09 .17 .19 .14
.77 .29 .16 .41 .79
DM
.95 .88 .85 .58
.99
.96 .64 -.19 -.47 -.37 -.11 .02 -.06
.30 -.09 .17 .00 .10 .40
LF
.86 .89 .62
.99
.69 -.19 -.50 -.40 -.12 .01 -.08 .94
.38 -.12 .21 .10 .21 .41 1.00
AF
-.38 -.67 -.44 -.06 -.05 -.10 .63 .68 .68 .73 .98
.99
.58 -.31 .26 .19 .46 .38 .95
%F
.57 .10 .04 .07 -.46 -.49 -.51 -.53 -.65
-.33 .43 -.12 .16 -.18 -.31 -.82 -.80 -1.01 NE
NE3 NE NE NE NE NE NE NE NE -.14 -.37 -.06 .02 .04 -.18 -.19 -.19 -.20 -.38
%P
%A
.07 .00 .06 -.36 -.40 -.39 -.41 -.43
-.42 .26 -.42 .02 -.40 -.60 -.80 -.76 -.82 NE .88
%M
.25 -.01 -.11 -.12 -.10 -.11 -.06
-.22 .31 -.24 .37 -.79 -.53 -.55 -.52 -.71 NE .83 .77
%BM
.33 .02 .01 .00 -.01 -.05
-.13 .14 -.88 -.71 .41 .15 -.37 -.29 -.22 NE .23 .08 -.96
%TM
of variance and covariance calcula
3
NE = Nonestimable, due to negative variance component.
SW = Shank width; SL = shank length; BM, DM, and TM = breast, drum, and thigh muscles, respectively; LF and A F = leaf f %M = % fat, % ash, % protein, and % moisture, respectively; %BM, %TM, %DM, %LF, and %AF = muscle and fat traits expressed as natural logarithms of LF, AF, and %F, respectively.
2
.13 .32 .31 .41 .00 .00 .01 .00 .01 -.04 .05 .05 .14 -.04 -.05 .01 .02 .02
.34
SW
(rp) correlation estimates from sires components
The I"G are above and rp are below the diagonal.
.44 .33 .76 .64 .64 .25 .28 .28 -.11 -.17 -.14 .11 .00 -.05 .18 .20 .28 .33 .29
BW SW SL BM TM DM LF AF %F %A %P %M %BM %TM %DM %LF %AF LNLF LNAF LN%F
1
BW
Trait 2
TABLE 6. Genetic (TQ) and phenotypic
djournals.org/ at Ernst Mayr Library of the Museum Comp Zoology, Harvard University on April 20, 2015
...
»W
.19
SW
.15 21
SL
.08 .18 .19
BM
.09 .21 .24 .23
TM
.10 .20 .22 .20 .11
DM .22 .24 .24 .27 .27 .22
LF .20 .23 .22 .25 .25 .21 .01
AF .19 .24 .25 .27 .25 .24 .08 .05
%F
%P .25 .23 .25 .28 .29 .25 .27 .25 .33 NE
%A NE2 NE NE NE NE NE NE NE NE .30 .30 .29 .34 .34 .29 .37 .35 .42 NE .16
%M
.19 .29
.24 .23 .24 .22 .17 .21 .25 .24 .24 NE
%BM
...
.33 .41 .18
.28 .28 .19 .23 .25 .29 .30 .29 .34 NE
%TM
of variance and covariance from
com
1 SW = Shank width; SL = shank length; BM, DM, and TM = breast, drum, and thigh muscles, respectively; LF and AF = leaf %M = % fat, % ash, % protein, and % moisture, respectively; %BM, %TM, %DM, %LF, and %AF = muscle and fat traits expressed a natural logarithms of LF, AF, and %F, respectively. 2 NE = Nonestimable, due to negative variance component.
%P %M %3M %TM %DM bLF %AF LNLF LNAF LN%F
%A
SL BM TM DM LF AF %F
bW
BW
Trait 1
TABLE 7. Standard errors of the genetic correlation (rQ) estimates from sire components
ournals.org/ at Ernst Mayr Library of the Museum Comp Zoology, Harvard University on April 20, 2015
1398
HAVENSTEIN ET AL.
Genetic correlation estimates of muscle weight measures with fat measures tended to be positive and low to moderate in M data (Table 4), but showed no definite trend in F data (Table 3). Muscle weights tended to be negatively associated with %M and %P. High positive rG were present among all the fat measures in both data sets (Tables 3 and 4). This suggests that the different fat measures were measuring the same genetic trait. However, it should be cautioned that the distributions of data for AF and LF were not normal, so some of these estimates should be viewed with caution. Genetic correlations of fat measures with %P and %M tended to be negative. Genetic correlation estimates for %BM, %TM, and %DM with the other traits tended to be erratic. This result is related to these traits having very low levels of additive genetic variance or very small estimates for the sire components of variance and covariance. It is not unusual for estimates of genetic correlations to behave in this manner when the estimates of the genetic parameters are small. REFERENCES Arthur, J. A., and H. Abplanalp, 1975. Linear estimates of heritability and genetic correlation for egg production, body weight, conformation and egg weight of turkeys. Poultry Sci. 54:11-23. Bartlett, M. S., 1937. Properties of sufficiency and statistical tests. Proc. R. Soc. Lond., Series A. 160:268-282. Bacon, W. L., K. E. Nestor, and P. A. Renner, 1986. The influence of genetic increases in body weight and shank width on the abdominal fat and carcass composition of turkeys. Poultry Sci. 65:391-397. Becker, W. A., 1978. Genotypic and phenotypic relations of abdominal fat in chickens. Pages 97-129 in: Proc. of the 27th National Breeder's Roundtable. Kansas City, MO. Becker, W. A., J. V. Spencer, L. W. Mirosh, and J. A. Verstrate, 1984. Genetic variation of abdominal fat, body weight, and carcass weight in a female broiler line. Poultry Sci. 63:607-611. Brown, D. R., L. L. Southern, and D. H. Baker, 1985. A comparison of methods for organ weight data adjustment in chicks. Poultry Sci. 64:366-369. Chambers, J. R., J. S. Gavora, and A. Fortin, 1981. Genetic changes in meat-type chickens in the last 20 years. Can. J. Anim. Sci. 61:555-563. Delabrosse, V., M. Dounaire, and J. Mallard, 1986. Les parametres genetiques de la composition corporelle chez la dinde. Pages 171-175 in: Proc. 7th Eur. Poult. Conf., Vol. 2, Paris, France, October 24-28. Fry, J. L., O. S. Rao, and L. D. Rasplicka, 1962. Factors affecting the yield of turkey parts. Poultry Sci. 41:1299-1303. Griffin, H. D., and C. C. Whitehead, 1985. Identification of lean or fat turkeys by measurement of plasma very
Downloaded from http://ps.oxfordjournals.org/ at Ernst Mayr Library of the Museum Comp Zoology, Harvard University on April 20, 2015
and DM weights were also positive and relatively low in males. The BM was, however, more highly correlated with BW than was TM or DM in both data sets (Tables 3 and 4). The existence of these relatively low rG demonstrates, as was pointed out by Havenstein et al. (1988), the need to take into account some characteristic of skeletal growth when selecting for growth rate in turkeys. The low rG between SW and SL (.10 in F and .32 in M) indicates that both bone length and width needs to be monitored in order to effect desirable changes in skeletal development. Shank width tended to show high (range = -.52 to -.90) negative correlations with all measures of fatness in the F data set, but consistently low positive correlations (range = .06 to .21) with all fat measures in the M data set. This result may explain why Nestor et al. (1985, 1987) reported significant line x sex interactions when comparing the performance of SWselected vs. BW-selected growth lines of turkeys (both of which lines were originally derived from the RBC2 line used herein). However, SL was positively correlated genetically with all fat measures in both data sets. The inconsistency of the rG for SW from the within-sex analyses again emphasizes, as pointed out by Havenstein et al. (1988), that one may be dealing with different "traits" in the two sexes when slaughter is at a common age. These differences in rG may simply reflect the differences in physiological development between the two sexes at 16 wk of age. Genetic correlation estimates among the three muscle groups (i.e., BM, TM, and DM) were positive but moderate in both data sets. The TM and DM were more highly correlated genetically with each other than with BM. This result further substantiates the hypothesis that selection for a trait such as BW can change certain parts of the body at different rates than others due to the differences in rG among the different muscle groups and among the different muscle groups and BW. The relatively low correlations of TM and DM with BM are considered to be extremely important because they indicate that selection for breast conformation scores may lead to disproportionate changes in the amount of breast muscle vs. the amount of leg muscles. Such a disproportionate change in these two muscle groups is thought to be a contributing factor to leg weakness problems (Nestor et al., 1985, 1987).
GENETIC COMPOSITION PARAMETERS IN TURKEYS
Nestor, K. E., 1984. Genetics of growth and reproduction in the turkey. 9. Long term selection for increased 16-wk body weight. Poultry Sci. 63:2114-2122. Nestor, K. E., W. L. Bacon, P. D. Moorhead, Y. M. Saif, G. B. Havenstein, and P. A. Renner, 1987. Comparison of bone and muscle growth in turkey lines selected for increased body weight and increased shank width. Poultry Sci. 66:1421-1428. Nestor, K. E., W. L. Bacon, Y. M. Saif, and P. A. Renner, 1985. The influence of genetic increases in shank width on body weight, walking ability, and reproduction of turkeys. Poultry Sci. 64:2248-2255. Neter, J., and W. Wasserman, 1974. Applied Linear Statistical Models. RichardD. Irwin, Inc., Homewood, IL. Nordstrom, J. O., R. H. Towner, G. B. Havenstein, and G. L. Walker, 1978. Influence of genetic strain, sex, and dietary energy level on abdominal fat deposition in broilers. Poultry Sci. 57:1176. (Abstr.) Proudman, J. A., W. J. Mellen, andD. L. Anderson, 1970. Utilization of feed in fast- and slow-growing lines of chickens. Poultry Sci. 49:961. (Abstr.) Ricard, F. H., and R. Rouvier, 1967. Etude de la composition anatomique du poulet. I. Variabilite de la repartition des parties corporelles chez des coquelets "Bresse Pile". Ann. Zootech. (Paris) 16:23-39. Ricard, F. H., and R. Rouvier, 1969. Etude de la composition anatomique du poulet. III. Variabilite de la repartition des parties corporelles dans une souche de type Cornish. Ann. Genet. Sel. Anim. 1:151-165. Siegel, P. B., 1984. Factors influencing fat deposition in meat poultry. 1. Genetics. Pages 51-52 in: Proc. XVIIth World's Poult. Congress, Helsinki, Finland. Swiger, L. A., W. R. Harvey, D. O. Everson, and K. E. Gregory, 1964. The variance of intraclass correlation involving groups with one observation. Biometrics 20:818-826. Tallis, G. M., 1959. Sampling errors of genetic correlation coefficients calculated from analyses of variance and covariance. Aust. J. Stat. 1:35-43. Wethli, E., and J.P.H. Wessels, 1973. The association between body fat content and thyroid activity, feed intake, mass gain, feed conversion and final body mass in growing chickens. Agroanamalia 5:83-88.
Downloaded from http://ps.oxfordjournals.org/ at Ernst Mayr Library of the Museum Comp Zoology, Harvard University on April 20, 2015
low density lipoprotein concentration. Br. Poult. Sci. 26:51-56. Gunsett, F. C , 1984. Linear index selection to improve traits defined as ratios. J. Anim. Sci. 59:1185-1193. Harshaw, H. M., andR. R. Rector, 1940. The composition of turkeys as affected by age and sex. Poultry Sci. 19:404-411. Hartung, T. E., and G. W. Froning, 1968. Variation of physical components of turkey carcasses as influenced by sex, age and strain. Poultry Sci. 47:1348-1355. Harvey, W. R., 1986. User's guide for LSMLMW. The Ohio State Univ., Columbus, OH. Havenstein, G. B., K. E. Nestor, V. D. Toelle, and W. L. Bacon, 1988. Estimates of genetic parameters in turkeys. 1. Body weight and skeletal characteristics. Poultry Sci. 67:1378-1387. Johnson, A. S., and R. S. Gowe, 1962. Modification of the growth pattern of the domestic turkey by selection at two ages. Pages 57-63 in: Proc. 12th World's Poult. Cong., Sydney, Australia. Leenstra, F. R., 1986. Effect of age, sex, genotype and environment on fat deposition in broiler chickens-a review. World's Poult. Sci. J. 42:12-15. Leenstra, F. R., and R. Pit, 1987. Fat deposition in a broiler sire strain. 2. Comparisons among lines selected for less abdominal fat, lower feed conversion ratio, and higher body weight after restricted and ad libitum feeding. Poultry Sci. 66:193-202. Leenstra, F. R., P.F.G. Vereijken, and R. Pit, 1986. Fat deposition in a broiler sire strain. 1. Phenotypic and genetic variation in abdominal fat, body weight and feed conversion. Poultry Sci. 65:1225-1235. Leeson, S., and J. D. Summers, 1980. Production and carcass characteristics of the large white turkey. Poultry Sci. 59:1237-1245. Lin, C. Y., G. W. Friars, and E. T. Moran, 1980. Genetic and environmental aspects of obesity in broilers. World's Poult. Sci. J. 36:103-111. Marks, H. L., 1979. Growth rate and feed efficiency of selected and non-selected broilers. Growth 43:80-90. Nestor, K. E., 1982. The influence of genetic increases in body weight on the abdominal fat pad of turkeys. Poultry Sci. 61:2301-2304.
1399
(Received for publication January 19, 1988) ABSTRACT Data were taken from 1,088 individuals-504 females (F) and 584 males (M)-produced by 34 sires and 168 dams from a randombred control line (RBC2) of turkeys maintained at the Ohio Agricultural Research and Development Center, Wooster, OH. Turkeys were used to estimate genetic and phenotypic parameters among 16-wk body weight and a number of body composition characteristics. Heritabilities (h2) were estimated from sire components of variance for 16-wk weights of muscles of breast (BM: F = .08, M = .35); thigh (TM: F = .12, M = .17); and drum (DM: F = .30, M = .44); weights of leaf fat (LF: F = .13, M = .51) and total abdominal fat (AF: F = .21, M = .55); and total percentage carcass fat (%F: F = .06, M = .24), percentage protein (F = .07, M = .20), and percent moisture (F = nonestimable, M = .11). The h2 of percentage carcass ash was nonestimable from both data sets due to negative estimates of the sire variance components. Due to the non-normal distributions of some of the fat measures, the fat measure data were converted to natural logarithms. Genetic estimates from the transformed data were similar to those from the nontransformed data. Genetic correlation (rG) estimates indicated a positive genetic association between body weight (BW) and all the fat characteristics measured. The BW was highly correlated with BM (F = .48, M = .86), TM (F = .81, M = 1.01), and DM (F = 1.04, M = .70). However, BW was correlated to a lesser degree with LF (F = .02, M = .27), AF (F = .25, M = .32), and %F (F = .93, M = .50). The BM was only moderately correlated with TM (F = .43, M = .28) and DM (F = .73, M = .27). The TM and DM were highly correlated (F = 1.02, M = .84). The existence of these correlations, which are less than unity, indicate that selection for BW alone will lead to increasing levels of fat in commercial turkeys, and to disproportionate increases in breast vs. leg muscles. (Key words: turkeys, body weight, heritability, genetic correlation, carcass composition) 1988 Poultry Science 67:1388-1399 INTRODUCTION
The goal of any food production enterprise is to produce consumer acceptable products of high quality at the most competitive price. Over the years, breeders of meat type poultry have concentrated their genetic improvement efforts on selection for heavier body weights and wider breast conformation at a given market weight. Due to the highly heritable nature of growth rate, their efforts have resulted in vastly improved growth rates for turkey and broiler populations. The effectiveness of their efforts on the growth rate of turkeys has been well documented in a review by Arthur and Abplanalp (1975), and in recent reports by Nestor (1984) and by Delabrosse et al. (1986).
'Salaries and research support provided by state and federal funds appropriated to the Ohio Agricultural Research and Development Center, The Ohio State University. Journal Article Number 3-88.
Traditionally, changes brought about in growth rates of chickens and turkeys have been accomplished without paying much attention to body composition. It is now well documented that increased growth rate has been accompanied by undesirable increases in body fat (for reviews, see Lin et al., 1980; Leenstra, 1986). Numerous workers have also shown that the genetic correlation between body weight and fat deposition, at least in broilers, is positive (Ricard and Rouvier, 1967, 1969; Proudman et al., 1970; Wethli and Wessels, 1973; Becker, 1978; Becker et al., 1984) and that selection towards increased growth rate has resulted in excessive amounts of fat at slaughter age in commercial broilers (Nordstrom et al., 1978; Marks, 1979; Chambers et al., 1981; Becker et al., 1984; Siegel, 1984; Leenstra et al., 1986; Leenstra and Pit, 1987). Some of these studies have also shown that different genetic stocks differ in the amount of fat present at market weight.
1388
Downloaded from http://ps.oxfordjournals.org/ at Ernst Mayr Library of the Museum Comp Zoology, Harvard University on April 20, 2015
Department of Poultry Science, The Ohio State University, Ohio Agricultural Research and Development Center, Wooster, Ohio 44691
GENETIC COMPOSITION PARAMETERS IN TURKEYS
MATERIALS AND METHODS
Data were collected at 16 wk of age on 1,088 fully pedigreed turkeys (504 females and 584 males) of a randombred control line (RBC2) maintained at the Ohio Agricultural Research and Development Center, Wooster, OH. The data reported herein are part of a larger study that also included measurements on a number of skeletal characteristics. The analyses of those data were reported in the first paper in this series (Havenstein et al., 1988), which also included a complete description of the materials and methods. As reported in the companion paper, one-half of the carcass was used for whole body analyses
of percentage carcass fat (%F), percentage ash (%A), percentage protein (%P), and percentage moisture (%M). The other half of the carcass was cut up and the breast muscles (BM), thigh muscles (TM), and drum muscles (DM) were weighed. Leaf fat (LF) and total abdominal fat (AF) were also weighed and recorded. Before analysis, muscle weights were multiplied by two, and they and AF and LF were used to calculate percentages of total live weight (%BM, %TM, %DM, %LF, and %AF). Thus, analyses of fat and of muscles were made on both a weight and a percentage of live weight basis. Sixteen-week body weight (BW), shank width (SW), and shank length (SL) were reported in the companion paper (Havenstein et al., 1988), but are also included herein as a common reference point. Because a preliminary analysis indicated that AF and LF measurements were not normally distributed, those measurements were transformed to natural logarithms before analysis. The transformed traits are referred to as LNAF and LNLF, respectively. The %F was also transformed in this manner and is reported as LN%F. Data were analyzed and variance and covariance components were estimated using the least-squares and maximum likelihood procedures outlined by Harvey (1986). The data set containing both sexes was analyzed using the following model: Yijklm = JJL + Si + d/Sij + Hk + SX, + bCXjjkim - x) + eijklm where: |x = the overall mean for trait Y; Sj = the random effect of the ith sire; d/Sy = the random effect of the j t h dam mated to the ith sire; Hk = the fixed effect of the kth hatch; SX, = the fixed effect of the 1th sex; b = the regression of Yyklm on the day of kill; Xyklm = the day of kill for the individual; x = the average day of kill; and eijkim = the random error associated with the measurement of each individual, which is assumed to be randomly and independently distributed with a mean of zero and a variance of a2. The data were also analyzed separately by sex using the same analysis as shown above except that sex was deleted from the model. This was done because turkeys have an extreme degree of sexual dimorphism for body, muscle, and fat weights, which results in large differences between means and variances for the two
Downloaded from http://ps.oxfordjournals.org/ at Ernst Mayr Library of the Museum Comp Zoology, Harvard University on April 20, 2015
Several investigators (Johnson and Gowe, 1962; Nestor, 1982; Bacon et al, 1986) have demonstrated that the amount of body fat has also increased with growth rate selection in turkeys, but as yet it is not considered to be a significant problem. However, there is increasing concern that current selection practices by the turkey breeding industry will eventually lead to the same consumer complaints about excessive fatness that the broiler industry has had to face over the past few years. Excessive fat levels in meat type poultry is a major concern. It leads both to economic losses for processors, who send much of the abdominal fat along with the viscera to the rendering plant, and to the consumer, who generally discards it. Large amounts of fat left in the carcass also lead to complaints and decreased consumer acceptance. It has been postulated (Nestor et al., 1985) that selection for increased growth rate has led to disproportionate increases in the relative amounts of breast vs. leg muscles, and that this disproportionate change has brought about at least part of the leg weakness problem that the turkey industry is currently facing. Only one study is available at this time on the genetic relationships among body weight, various body parts, and fat deposition in turkeys (Delabrosse et al., 1986). The purpose of the present study was to examine the inheritance of and the genetic and phenotypic relationships among live body weight and a number of muscle and fat measurements in turkeys, as well as genetic and phenotypic relationships of those traits with shank width and length. The latter traits appear to be a useful measure for the improvement of the skeletal support system (Nestor et al., 1987; Havenstein et al., 1988).
1389
HAVENSTEIN ET AL.
1390
RESULTS AND DISCUSSION
Testsfor Heterogeneity of the Variancesfrom the Within Sex Analyses. Phenotypic variances for the same' 'trait'' from the within sex analyses were found to be heterogeneous for all of the traits except %A, %M, %BM, %TM, %DM, and LN%F. Thus, the data from males (M) and females (F) for the majority of traits should be considered separately. However, because the ANOVA is a very robust technique, genetic parameter estimates were also made from the com-
bined-data set and are provided herein for comparison purposes. Means and Phenotypic Standard Deviations. The overall means and the phenotypic standard deviations from both within-sex and combineddata set analyses for BW, S W, SL, all the muscle and fat weights, all the muscle and fat percentages, and the transformation of fat measures to natural logarithms are presented in Table 1. The M had significantly higher (P<.05) mean values for BW, SW, SL, BM, TM, DM, %P, %A, and %M than F. However, F had significantly higher (P<.05) mean values for all fat measures and muscle traits when expressed as a percentage of body weight. These observations are consistent with results of most carcass composition studies, where M generally have larger absolute values for body and muscle weights, but F have larger values for fat measures and percentage muscle (Harshaw and Rector, 1940; Fry et al., 1962; Hartung and Froning, 1968; Leeson and Summers, 1980). Heritability Estimates. The h 2 estimates for BW, SW, SL, and the carcass characteristics
TABLE 1. Overall means and standard deviations (op) for traits studied as calculated within sex and with sexes combined Females
Males Trait 1 BW, kg SW, mm SL, cm
op
X
7.34
.56
Sexes combined op
X
5.25
op
X
.42
6.37
.50
13.3 15.6
.5 .5
11.8 12.8
.5 .5
12.6 14.3
.5 .5
665.0 347.2 260.7
73.2 41.2 29.8
489.0 262.9 200.3
56.8 32.6 22.8
583.5 308.1 232.7
66.1 37.5 26.8
3.0 6.5
3.1 5.4
7.9
7.2
5.2
15.4
12.7
10.6
5.4 9.6
%F %A %P %M
13.0 12.6 68.0 70.8
3.0 2.5 3.5 1.8
17.9 11.8 63.9 69.0
4.2 2.5 4.3 1.8
15.3 12.2 66.1 70.0
3.6 2.5 3.9 1.8
%BM %TM %DM
18.1
1.2 .9 .6
18.6 10.0
1.5 .9 .6
18.3
1.4 .9 .6
BM, g TM, g DM, g LF,g AF,g
%LF %AF
LNLF LNAF LN%F
9.5 7.1
7.6
9.7 7.3
.04 .09
.04 .07
.15 .29
.13 .23
.09 .18
.09 .16
1.17 1.84 2.61
.59 .56 .22
1.89 2.51 2.91
.76 .76 .22
1.50 2.15 2.75
.68 .66 .22
1 SW = Shank width; SL = shank length; BM, DM, and TM = breast, drum, and thigh muscles, respectively; LF and AF = leaf fat and total abdominal fat, respectively; %F, %A, %P, and %M = % fat, % ash, % protein, and % moisture, respectively; %BM, %TM, %DM, %LF, and %AF = muscle and fat traits expressed as a % of live BW, respectively; LNLF, LNAF, and LN%F = natural logarithms of LF, AF, and %F, respectively.
Downloaded from http://ps.oxfordjournals.org/ at Ernst Mayr Library of the Museum Comp Zoology, Harvard University on April 20, 2015
sexes. Phenotypic variances for the same "trait" from these within sex analyses were tested for heterogeneity using the method of Bartlett (1937) as shown by Neter and Wasserman (1974). The results are reported from both the within-sex and the pooled-data set analyses. Standard errors for the heritability (h2) and genetic correlation (rG) estimates were calculated by modifications of the methods of Tallis (1959) and Swiger et al. (1964) as described by Harvey (1986).
GENETIC COMPOSITION PARAMETERS IN TURKEYS
culated separately for the two sexes, but differences between their estimates tended to be in the opposite direction to those reported herein. They, however, slaughtered M and F at different ages, whereas M and F were slaughtered at the same ages in this study. This difference in procedure, and in the estimates obtained, indicates that differences in physiological development at a common killing age may affect the expression of the "same trait" differently in the two sexes. If so, then a "trait" measured at the same chronological age in the two sexes may actually be two different traits, and therefore, should be included separately in genetic selection programs so as to maximize the improvement of overall genetic merit. The ex2 h2 estimates for BM, TM, and DM were higher than those for their comparable percentages of BW values, i.e., %BM, %TM, and %DM. Fat characteristics transformed to natural logarithms also had higher h 2 estimates from the
TABLE 2. Heritability estimates (h2) and their standard errors (SE) from sire (CT|) and dam (afi) components of variance when measured within sex and with sexes combined Sexes c o m b i n e d
Females
Males
°s
°s
°y
°d
"s
fd
Traits 1
h2
SE
h2
SE
h2
SE
h2
SE
h2
SE
h2
BW
.60
.19
.63
17
.23
.14
1.22
.20
.45
.13
.82
.12
SW SL
.47 .54
.17 .19
.48 .58
17 17
.55 .43
.20 .18
.68 .06
.20 .18
.46 .51
.14 .14
.53 .27
.11 .10
BM TM DM
.35 .17 .44
.15 .11 .17
.87 .77 .39
18 18 17
.08 .12 .30
.10 .11 .15
.91 .80 .84
.20 .20 .20
.25 .21 .35
.09 .08 .11
.86 .64 .49
.13 .12 .11
LF AF
.51 .55
.18 .19
.57 .61
17 17
.13 .21
.11 .13
.78 .92
.20 .20
.21 .27
.08 .10
.55 .58
.11 .11
%F %A %P %M
.24 NE2 .20 .11
.12
17 16 16 15
.06 NE .07 NE
.10
.68 .18 .61 .80
.20 .18 .20 .20
.16 NE .16 .08
.07
.12 .09
.60 .19 .31 .08
.07 .05
.46 .05 .21 .13
.11 .08 .09 .09
%BM %TM %DM
.23 .00 .36
.12 .07 .15
.72 .68 .38
18 18 17
.19 .07 .04
.13 .10 .09
.20 .48 .50
.19 .19 .19
.19 .11 .22
.08 .06 .09
.44 .22 .27
.11 .09 .10
%LF %AF
.51 .56
.18 .19
.59 .64
17 17
.14 .21
.12 .13
.71 .87
.20 .20
.18 .23
.08 .09
.48 .51
.11 .11
LNLF LNAF
.64 .63
.20 .20
.41 .48
17 17
.20 .28
.13 .15
.68 .81
.20 .20
.38 .41
.12 .13
.55 .60
.11 .11
LN%F
.25
.13
.56
17
.04
.09
.70
.20
.18
.08
.45
.11
.10
1
SE
SW = Shank width; SL = shank length; BM, DM, and TM = breast, drum, and thigh muscles, respectively; LF and AF = leaf fat and total abdominal fat, respectively; %F, %A, %P, and %M = % fat, % ash, % protein, and % moisture, respectively; %BM, %TM, %DM, %LF, and %AF = muscle and fat traits expressed as a % of live BW, respectively; LNLF, LNAF, and LN%F = natural logarithms of LF, AF, and %F, respectively. 2
NE = Nonestimable, due to a negative variance component estimate.
Downloaded from http://ps.oxfordjournals.org/ at Ernst Mayr Library of the Museum Comp Zoology, Harvard University on April 20, 2015
are summarized in Table 2. The BW, SW, and SL estimates were reported in the companion paper (Havenstein et al., 1988), but are also reported herein because of their overall importance to turkey breeding programs, and to show their relationships with the traits in this portion of the study. The h 2 estimates for the muscle and fat characteristics show a similar pattern to those for skeletal characteristics reported by Havenstein et al. (1988), in that the estimates from sire components of variance (a 2 ) from M data are generally higher than those from F data. However, h estimates from dam components of variance (a2,) were higher in F than in M data. These results support the conclusion by Havenstein et al. (1988) that the expression of genetic variation for the same trait at the same killing age is influenced by the sex of the individuals involved. However, Delabrosse et al. (1986) also reported h 2 estimates that were cal-
1391
1392
HAVENSTEIN ET AL.
ficients of variation for fat measures in the range of 25 to 30%, whereas the turkey data reported herein show coefficients of variation for LF and AF approaching 100% (see the means and standard errors in Table 1). The question is, which of these measures of fatness provides the best criterion for selection purposes? It appears from this study that %F provides the best information from a statistical point of view, because it is normally distributed, and it also has the highest correlation with BW. However, from an application point of view, as measurement of %F requires the bird to be sacrificed and a complete carcass analysis performed, it would be difficult and expensive to use %F as a routine selection criterion for a commercial turkey breeding program. It also appears that in spite of the non-normality of the distributions for LF and AF, that LF, AF, and %F were highly correlated genetically (r G > .95), and that they are, therefore, equally good measures of fatness. However, because all of the measures used in this study require the use of sacrificed sibs to obtain the selection data, it appears that until a better measure is found, selection for reduced levels of the very low density lipoproteins as suggested by Griffin and Whitehead (1985) is the method of choice for reducing fatness in turkeys. Griffin and Whitehead's method may be less costly to use, and probably provides a better criterion for selection than any of the measures reported in this study. It is certain, however, based on the rG present in these data and in the data of Delabrosse et al. (1986), that continued selection for BW and breast conformation score in turkeys, a measure of total breast muscle, will eventually result in excessively fat turkeys, just as it has with broiler chickens. The other whole body analyses, i.e., for %P, %A, and %M, also showed rG with BW (Tables 3 , 4 , and 6). Negative of estimates from both the M and F data prevented the actual calculation of rG estimates for %A. The r P for these two traits was -.11 for both sexes. All estimates of the rG between BW and %P and %M were negative, as were the rP between these traits. Again, a negative of from the F data prevented the calculation of an estimate for %M from that data set. Nevertheless, it appears that as BW is increased by selection, both %P and %M will decrease. Estimates of the rG between SW and BW were positive and relatively low (Havenstein et al, 1988). The rG between SW and BM, TM,
Downloaded from http://ps.oxfordjournals.org/ at Ernst Mayr Library of the Museum Comp Zoology, Harvard University on April 20, 2015
of than did their comparable nontransformed measurements. These h patterns were observed in estimates from both sexes (Table 2). Estimates of Genetic (rG) andPhenotypic (rP) Correlations. Summaries of the rG and r P correlations, calculated from the sire components of variance and covariance from the within-sex analyses, can be found in Tables 3 and 4. Standard errors of rG estimates in Tables 3 and 4 are presented in Table 5. The rG and rP estimates from the combined-data set analysis are provided in Table 6, and standard errors for the rG estimates from the combined analysis are given in Table 7. The rG estimates of BW with BM, TM, and DM weights tended to be high (range .48 to 1.04) in both sexes (Tables 3 and 4). The TM and DM weights of F appeared to be somewhat more highly correlated with BW than the BM weights. Estimates from these same muscle weight groups, when expressed as a percentage of live weight, are far less clear cut. For example, the rG of BW with %BM and %TM were negative in both sexes, whereas correlations with %DM were negative from the M data and highly positive from the F data. Extreme caution should be used in evaluating and using correlation estimates that are based on a ratio trait, especially one which is being correlated with a trait that is one of its component parts. Gunsett (1984) and Brown et al. (1985) discussed some of the inherent problems associated with using ratio or percentage traits in which the numerator and denominator are moderately to highly correlated. Genetic correlation estimates of B W with all fat measures were positive in both sexes (Tables 3 and 4). The LF showed the lowest rG with BW (F = .02, M = .27). Total AF was slightly (F = .25, M = .32), and %F was considerably more highly correlated (F = .93, M = .50) with BW than was LF. Fat measures were also expressed as a percentage of body weight, and as natural logarithm transformations of the original measures. The rG of BW with the transformed fat measures were very similar to those from their comparable nontransformed measures (Tables 3 , 4 , and 6). Transformation to natural logarithms was done because LF and AF are not normally distributed and are extremely skewed in the upward direction. Becker et al. (1984) reported a similar situation in broilers, but the non-normality and skewness is even greater in the present turkey data than in the broiler data. For example, Becker et al. (1984) reported coef-
GENETIC COMPOSITION PARAMETERS IN TURKEYS
o o CO 00 ^
^+
1
rt
" C\l
o
z
1 Z
Z
w 1 z
-
1
o in o
in
00
o
NO W
fN} o 00 fN] 1 1
1
ON O 00
o
o *t o
o o
1
1
r~
00 NO 00 ON O N m
t
~
CO NO NO CO CO ON O N
o o
00 ON 1 ^ O ON O
1 z
Z
1
1
rt^
1
ON
w Z
~
ON
1 1
Z
O l " ~ f ^ - O © O 0 \ O < ' - I f • rt • rA W I I 1 Z i *o t-> H r n CO 0 y M 00 • n rt O : i i i Z 1
O O O N O f i N C O N N "* f f rt U <-i U] 1 I Z Z r^so^O"%'Hr^GOoo^ f
f
a*
l ' I H M M U r i [I] I I I Z Z
I
Z Z Z Z Z Z Z Z Z Z Z
I
I
I I
U U U t i U U K U U Z Z Z Z Z Z Z Z Z
I I I
I _
U, tin U.
S S j S S S t t . n . t < a . S e o H Q J < Z Z Z
Downloaded from http://ps.oxfordjournals.org/ at Ernst Mayr Library of the Museum Comp Zoology, Harvard University on April 20, 2015
00 (N C-l NO
NO O N CO
7
.35 -.60
-H
.29 .24 .24 -.01 .02 -.03 .10 -.01 -.10 .10 .03 .03 -.03 -.01 .00 .03 -.03
.53 .32
SL
.63 .44 .10 .13 .20 -.11 -.09 -.03 .73 .15 -.10 .05 .06 .12 .15 .19
.86 .41 .34
BM
.56 .14 .16 .15 -.07 -.08 -.07 .32 .76 .19 .10 .11 .14 .17 .15
1.01 .29 -.07 .28
TM
.17 .18 .16 -.05 -.08 -.03 .04 .22 .75 .13 .14 .18 .20 .16
.70 .30 .00 .27 .84
DM
.95 .57 -.12 -.45 -.25 -.09 -.01 .02 .99 .95 .93 .88 .54
.27 .21 .05 .16 .35 .40
LF
.62 -.14 -.47 -.26 -.10 -.02 .00 .94 .99 .88 .93 .58
.32 .21 .05 .19 .41 .39 1.00
AF
-.19 -.36 -.04 .01 .02 -.11 -.13 -.13 -.16 -.39 .42 .07 .06 .07 -.44 -.46 -.45 -.48 -.59
-.23 .10 .04 .06 -.26 -.49 -.92 -.98 -1.16 NE
NE3 NE NE NE NE NE NE NE NE
.50 .06 .34 .39 .62 .34 1.00 1.03 -.40 -.60 -.26 -.04 -.07 -.06 .55 .61 .59 .66 .99
%P
%A
%F
(rp) correlation estimates from sire components %BM
-.25 -.11 -.27 .18 -.55 -.30 .42 -.09 -.43 -1.30 -.58 -.76 -.31 -.16 -.35 -.17 -.47 -.09 NE NE .47 .08 .30 .06 -.01 .28 .04 -.09 -.25 -.10 -.25 -.11 -.26 -.09 -.26 -.11 -.26 -.05
%M
.30 -.0 -.02 -.0 -.04 -.08
-1.84 -1.54 -7.50 -7.96 -2.04 .14 .33 .40 .65 NE -.04 -1.48 -12.82
%TM
of variance and covar
3
N E = Nonestimable, due to negative variance component.
SW = Shank width; SL = shank length; BM, DM, and TM = breast, drum, and thigh muscles, respectively; LF and AF = leaf f %M = % fat, % ash, % protein, and % moisture, respectively; %BM, %TM, %DM, %LF, and %AF = muscle and fat traits expressed a natural logarithms of LF, AF, and %F, respectively.
3
.09 .28 .27 .40 .05 .06 .07 .03 -.06 -.11 .02 .02 .18 .03 .02 .03 .04 .07
.33
SW
The TQ are above and rp are below the diagonal.
.39 .32 .79 .62 .61 .23 .28 .32 -.11 -.20 -.09 .16 -.03 -.07 .17 .20 .25 .31 .32
BW SW SL BM TM DM LF AF %F %A %P %M %BM %TM %DM %LF %AF LNLF LNAF LN%F
1
BW
Trait 2
TABLE 4. Genetic (re) and phenotypic
djournals.org/ at Ernst Mayr Library of the Museum Comp Zoology, Harvard University on April 20, 2015
.23 .31 .25 .24 .24 .27 NE .30 .36 .27 NE .21 .24 .24 .23 .23 .26
.19 .26
SL
.30 .25 .26 .25 .27 NE .32 .39 .24 NE .25 .26 .26 .25 .24 .27
.37 .36 .41
BM
.15 .29 .28 .31 NE .39 .48 .22 NE .33 .30 .29 .29 .27 .30
.19 .23 .37 .49
TM
.22 .22 .27 NE .29 .36 .23 NE .15 .23 .23 .22 .21 .26
.08 .23 .25 .32 .15
DM
.01 .09 NE .30 .39 .29 NE .25 .00 .01 .01 .01 .09
.48 .38 .39 .67 .49 .42
LF
.07 NE .29 .38 .29 NE .25 .01 .00 .01 .01 .08
.36 .30 .30 .50 .45 .34 .04
AF
NE .44 .48 .34 NE .31 .09 .08 .08 .07 .00
.68 .78 .75 1.22 .79 .74 .43 .62
%F
.45 .33 NE .30 .30 .29 .28 .28 .43
1.07 1.04 .71 1.49 .90 .92 1.68 1.52 NE NE
NE3 NE NE NE NE NE NE NE NE NE NE NE NE NE NE NE NE NE NE
%P
%A
.43 NE .37 .39 .38 .37 .37 .47
NE NE NE NE NE NE NE NE NE NE NE
%M
NE .24 .29 .29 .28 .28 .34
.35 .29 .33 .46 .41 .33 .58 .41 1.86 NE 1.76 NE
%BM
NE NE NE NE NE NE
.52 .41 .47 .73 .41 .46 .67 .55 1.32 NE 1.76 NE .53
%TM
NE = Nonestimable, due to negative variance component.
''NP = Not presented because of standard errors > 2.00 due to sire components near zero.
3
SW = Shank width; SL = shank length; BM, DM, and TM = breast, drum, and thigh muscles, respectively; LF and AF = leaf %M = % fat, % ash, % protein, and % moisture, respectively; %BM, %TM, %DM, %LF, and %AF = muscle-and fat traits expressed a natural logarithms of LF, AF, and %F, respectively.
2
.23 .23 .29 .23 .24 .24 .29 NE .31 .38 .29 NE .26 .24 .24 .24 .24 .29
.24
SW
of variance and covariance for data on fem
Standard errors for the TQ estimates from the female data {Table 3) are above the diagonal, and for the TQ estimates from the
.21 .18 .07 .11 .13 .22 .21 .22 NE .30 .36 .28 NE .25 .23 .22 .21 .20 .21
BW SW SL BM TM DM LF AF %F %A %P %M %BM %TM %DM %LF %AF LNLF LNAF LN%F
1
BW
Trait 2
TABLE 5. Standard errors of genetic correlation (rQ) estimates from sire components
ournals.org/ at Ernst Mayr Library of the Museum Comp Zoology, Harvard University on April 20, 2015
.25 .22 .26 .02 .04 -.05 .07 .01 -.03 .03 .00 .06 -.01 .01 -.02 .01 -.06
.60 .18
SL
.62 .47 .10 .12 .15 -.11 -.05 -.05 .72 .17 -.05 .06 .07 .13 .15 .16
.82 .50 .44
BM
.59 .18 .19 .14 -.06 -.08 -.09 .26 .76 .21 .13 .15 .18 .21 .15
.85 .38 .07 .34
TM
.13 .14 .12 -.05 -.06 -.05 .06 .24 .73 .09 .09 .17 .19 .14
.77 .29 .16 .41 .79
DM
.95 .88 .85 .58
.99
.96 .64 -.19 -.47 -.37 -.11 .02 -.06
.30 -.09 .17 .00 .10 .40
LF
.86 .89 .62
.99
.69 -.19 -.50 -.40 -.12 .01 -.08 .94
.38 -.12 .21 .10 .21 .41 1.00
AF
-.38 -.67 -.44 -.06 -.05 -.10 .63 .68 .68 .73 .98
.99
.58 -.31 .26 .19 .46 .38 .95
%F
.57 .10 .04 .07 -.46 -.49 -.51 -.53 -.65
-.33 .43 -.12 .16 -.18 -.31 -.82 -.80 -1.01 NE
NE3 NE NE NE NE NE NE NE NE -.14 -.37 -.06 .02 .04 -.18 -.19 -.19 -.20 -.38
%P
%A
.07 .00 .06 -.36 -.40 -.39 -.41 -.43
-.42 .26 -.42 .02 -.40 -.60 -.80 -.76 -.82 NE .88
%M
.25 -.01 -.11 -.12 -.10 -.11 -.06
-.22 .31 -.24 .37 -.79 -.53 -.55 -.52 -.71 NE .83 .77
%BM
.33 .02 .01 .00 -.01 -.05
-.13 .14 -.88 -.71 .41 .15 -.37 -.29 -.22 NE .23 .08 -.96
%TM
of variance and covariance calcula
3
NE = Nonestimable, due to negative variance component.
SW = Shank width; SL = shank length; BM, DM, and TM = breast, drum, and thigh muscles, respectively; LF and A F = leaf f %M = % fat, % ash, % protein, and % moisture, respectively; %BM, %TM, %DM, %LF, and %AF = muscle and fat traits expressed as natural logarithms of LF, AF, and %F, respectively.
2
.13 .32 .31 .41 .00 .00 .01 .00 .01 -.04 .05 .05 .14 -.04 -.05 .01 .02 .02
.34
SW
(rp) correlation estimates from sires components
The I"G are above and rp are below the diagonal.
.44 .33 .76 .64 .64 .25 .28 .28 -.11 -.17 -.14 .11 .00 -.05 .18 .20 .28 .33 .29
BW SW SL BM TM DM LF AF %F %A %P %M %BM %TM %DM %LF %AF LNLF LNAF LN%F
1
BW
Trait 2
TABLE 6. Genetic (TQ) and phenotypic
djournals.org/ at Ernst Mayr Library of the Museum Comp Zoology, Harvard University on April 20, 2015
...
»W
.19
SW
.15 21
SL
.08 .18 .19
BM
.09 .21 .24 .23
TM
.10 .20 .22 .20 .11
DM .22 .24 .24 .27 .27 .22
LF .20 .23 .22 .25 .25 .21 .01
AF .19 .24 .25 .27 .25 .24 .08 .05
%F
%P .25 .23 .25 .28 .29 .25 .27 .25 .33 NE
%A NE2 NE NE NE NE NE NE NE NE .30 .30 .29 .34 .34 .29 .37 .35 .42 NE .16
%M
.19 .29
.24 .23 .24 .22 .17 .21 .25 .24 .24 NE
%BM
...
.33 .41 .18
.28 .28 .19 .23 .25 .29 .30 .29 .34 NE
%TM
of variance and covariance from
com
1 SW = Shank width; SL = shank length; BM, DM, and TM = breast, drum, and thigh muscles, respectively; LF and AF = leaf %M = % fat, % ash, % protein, and % moisture, respectively; %BM, %TM, %DM, %LF, and %AF = muscle and fat traits expressed a natural logarithms of LF, AF, and %F, respectively. 2 NE = Nonestimable, due to negative variance component.
%P %M %3M %TM %DM bLF %AF LNLF LNAF LN%F
%A
SL BM TM DM LF AF %F
bW
BW
Trait 1
TABLE 7. Standard errors of the genetic correlation (rQ) estimates from sire components
ournals.org/ at Ernst Mayr Library of the Museum Comp Zoology, Harvard University on April 20, 2015
1398
HAVENSTEIN ET AL.
Genetic correlation estimates of muscle weight measures with fat measures tended to be positive and low to moderate in M data (Table 4), but showed no definite trend in F data (Table 3). Muscle weights tended to be negatively associated with %M and %P. High positive rG were present among all the fat measures in both data sets (Tables 3 and 4). This suggests that the different fat measures were measuring the same genetic trait. However, it should be cautioned that the distributions of data for AF and LF were not normal, so some of these estimates should be viewed with caution. Genetic correlations of fat measures with %P and %M tended to be negative. Genetic correlation estimates for %BM, %TM, and %DM with the other traits tended to be erratic. This result is related to these traits having very low levels of additive genetic variance or very small estimates for the sire components of variance and covariance. It is not unusual for estimates of genetic correlations to behave in this manner when the estimates of the genetic parameters are small. REFERENCES Arthur, J. A., and H. Abplanalp, 1975. Linear estimates of heritability and genetic correlation for egg production, body weight, conformation and egg weight of turkeys. Poultry Sci. 54:11-23. Bartlett, M. S., 1937. Properties of sufficiency and statistical tests. Proc. R. Soc. Lond., Series A. 160:268-282. Bacon, W. L., K. E. Nestor, and P. A. Renner, 1986. The influence of genetic increases in body weight and shank width on the abdominal fat and carcass composition of turkeys. Poultry Sci. 65:391-397. Becker, W. A., 1978. Genotypic and phenotypic relations of abdominal fat in chickens. Pages 97-129 in: Proc. of the 27th National Breeder's Roundtable. Kansas City, MO. Becker, W. A., J. V. Spencer, L. W. Mirosh, and J. A. Verstrate, 1984. Genetic variation of abdominal fat, body weight, and carcass weight in a female broiler line. Poultry Sci. 63:607-611. Brown, D. R., L. L. Southern, and D. H. Baker, 1985. A comparison of methods for organ weight data adjustment in chicks. Poultry Sci. 64:366-369. Chambers, J. R., J. S. Gavora, and A. Fortin, 1981. Genetic changes in meat-type chickens in the last 20 years. Can. J. Anim. Sci. 61:555-563. Delabrosse, V., M. Dounaire, and J. Mallard, 1986. Les parametres genetiques de la composition corporelle chez la dinde. Pages 171-175 in: Proc. 7th Eur. Poult. Conf., Vol. 2, Paris, France, October 24-28. Fry, J. L., O. S. Rao, and L. D. Rasplicka, 1962. Factors affecting the yield of turkey parts. Poultry Sci. 41:1299-1303. Griffin, H. D., and C. C. Whitehead, 1985. Identification of lean or fat turkeys by measurement of plasma very
Downloaded from http://ps.oxfordjournals.org/ at Ernst Mayr Library of the Museum Comp Zoology, Harvard University on April 20, 2015
and DM weights were also positive and relatively low in males. The BM was, however, more highly correlated with BW than was TM or DM in both data sets (Tables 3 and 4). The existence of these relatively low rG demonstrates, as was pointed out by Havenstein et al. (1988), the need to take into account some characteristic of skeletal growth when selecting for growth rate in turkeys. The low rG between SW and SL (.10 in F and .32 in M) indicates that both bone length and width needs to be monitored in order to effect desirable changes in skeletal development. Shank width tended to show high (range = -.52 to -.90) negative correlations with all measures of fatness in the F data set, but consistently low positive correlations (range = .06 to .21) with all fat measures in the M data set. This result may explain why Nestor et al. (1985, 1987) reported significant line x sex interactions when comparing the performance of SWselected vs. BW-selected growth lines of turkeys (both of which lines were originally derived from the RBC2 line used herein). However, SL was positively correlated genetically with all fat measures in both data sets. The inconsistency of the rG for SW from the within-sex analyses again emphasizes, as pointed out by Havenstein et al. (1988), that one may be dealing with different "traits" in the two sexes when slaughter is at a common age. These differences in rG may simply reflect the differences in physiological development between the two sexes at 16 wk of age. Genetic correlation estimates among the three muscle groups (i.e., BM, TM, and DM) were positive but moderate in both data sets. The TM and DM were more highly correlated genetically with each other than with BM. This result further substantiates the hypothesis that selection for a trait such as BW can change certain parts of the body at different rates than others due to the differences in rG among the different muscle groups and among the different muscle groups and BW. The relatively low correlations of TM and DM with BM are considered to be extremely important because they indicate that selection for breast conformation scores may lead to disproportionate changes in the amount of breast muscle vs. the amount of leg muscles. Such a disproportionate change in these two muscle groups is thought to be a contributing factor to leg weakness problems (Nestor et al., 1985, 1987).
GENETIC COMPOSITION PARAMETERS IN TURKEYS
Nestor, K. E., 1984. Genetics of growth and reproduction in the turkey. 9. Long term selection for increased 16-wk body weight. Poultry Sci. 63:2114-2122. Nestor, K. E., W. L. Bacon, P. D. Moorhead, Y. M. Saif, G. B. Havenstein, and P. A. Renner, 1987. Comparison of bone and muscle growth in turkey lines selected for increased body weight and increased shank width. Poultry Sci. 66:1421-1428. Nestor, K. E., W. L. Bacon, Y. M. Saif, and P. A. Renner, 1985. The influence of genetic increases in shank width on body weight, walking ability, and reproduction of turkeys. Poultry Sci. 64:2248-2255. Neter, J., and W. Wasserman, 1974. Applied Linear Statistical Models. RichardD. Irwin, Inc., Homewood, IL. Nordstrom, J. O., R. H. Towner, G. B. Havenstein, and G. L. Walker, 1978. Influence of genetic strain, sex, and dietary energy level on abdominal fat deposition in broilers. Poultry Sci. 57:1176. (Abstr.) Proudman, J. A., W. J. Mellen, andD. L. Anderson, 1970. Utilization of feed in fast- and slow-growing lines of chickens. Poultry Sci. 49:961. (Abstr.) Ricard, F. H., and R. Rouvier, 1967. Etude de la composition anatomique du poulet. I. Variabilite de la repartition des parties corporelles chez des coquelets "Bresse Pile". Ann. Zootech. (Paris) 16:23-39. Ricard, F. H., and R. Rouvier, 1969. Etude de la composition anatomique du poulet. III. Variabilite de la repartition des parties corporelles dans une souche de type Cornish. Ann. Genet. Sel. Anim. 1:151-165. Siegel, P. B., 1984. Factors influencing fat deposition in meat poultry. 1. Genetics. Pages 51-52 in: Proc. XVIIth World's Poult. Congress, Helsinki, Finland. Swiger, L. A., W. R. Harvey, D. O. Everson, and K. E. Gregory, 1964. The variance of intraclass correlation involving groups with one observation. Biometrics 20:818-826. Tallis, G. M., 1959. Sampling errors of genetic correlation coefficients calculated from analyses of variance and covariance. Aust. J. Stat. 1:35-43. Wethli, E., and J.P.H. Wessels, 1973. The association between body fat content and thyroid activity, feed intake, mass gain, feed conversion and final body mass in growing chickens. Agroanamalia 5:83-88.
Downloaded from http://ps.oxfordjournals.org/ at Ernst Mayr Library of the Museum Comp Zoology, Harvard University on April 20, 2015
low density lipoprotein concentration. Br. Poult. Sci. 26:51-56. Gunsett, F. C , 1984. Linear index selection to improve traits defined as ratios. J. Anim. Sci. 59:1185-1193. Harshaw, H. M., andR. R. Rector, 1940. The composition of turkeys as affected by age and sex. Poultry Sci. 19:404-411. Hartung, T. E., and G. W. Froning, 1968. Variation of physical components of turkey carcasses as influenced by sex, age and strain. Poultry Sci. 47:1348-1355. Harvey, W. R., 1986. User's guide for LSMLMW. The Ohio State Univ., Columbus, OH. Havenstein, G. B., K. E. Nestor, V. D. Toelle, and W. L. Bacon, 1988. Estimates of genetic parameters in turkeys. 1. Body weight and skeletal characteristics. Poultry Sci. 67:1378-1387. Johnson, A. S., and R. S. Gowe, 1962. Modification of the growth pattern of the domestic turkey by selection at two ages. Pages 57-63 in: Proc. 12th World's Poult. Cong., Sydney, Australia. Leenstra, F. R., 1986. Effect of age, sex, genotype and environment on fat deposition in broiler chickens-a review. World's Poult. Sci. J. 42:12-15. Leenstra, F. R., and R. Pit, 1987. Fat deposition in a broiler sire strain. 2. Comparisons among lines selected for less abdominal fat, lower feed conversion ratio, and higher body weight after restricted and ad libitum feeding. Poultry Sci. 66:193-202. Leenstra, F. R., P.F.G. Vereijken, and R. Pit, 1986. Fat deposition in a broiler sire strain. 1. Phenotypic and genetic variation in abdominal fat, body weight and feed conversion. Poultry Sci. 65:1225-1235. Leeson, S., and J. D. Summers, 1980. Production and carcass characteristics of the large white turkey. Poultry Sci. 59:1237-1245. Lin, C. Y., G. W. Friars, and E. T. Moran, 1980. Genetic and environmental aspects of obesity in broilers. World's Poult. Sci. J. 36:103-111. Marks, H. L., 1979. Growth rate and feed efficiency of selected and non-selected broilers. Growth 43:80-90. Nestor, K. E., 1982. The influence of genetic increases in body weight on the abdominal fat pad of turkeys. Poultry Sci. 61:2301-2304.
1399