- Email: [email protected]
Boar taint in pigs selected for components of efficient lean growth rate
Meat Science 54 (2000) 147±153
www.elsevier.com/locate/meatsci
Boar taint in pigs selected for components of ecient lean growth rate N.D. Cameron a,*, J.C. Penman a, A.C. Fisken a, G.R. Nute b, A.M. Perry b, F.W. Whittington b a Roslin Institute (Edinburgh), Roslin, Midlothian EH25 9PS, UK Division of Food Animal Science, School of Veterinary Science, University of Bristol, Langford BS4 05U, UK
b
Received 29 March 1999; accepted 3 July 1999
Abstract Genetic and nutritional eects on the boar taint traits of androstenone, skatole and indole and the eating quality traits associated with boar taint were examined by testing animals from four selection lines and a control line on isoenergetic diets, which diered in ileal digestible lysine: digestible energy (0.40, 0.76 and 1.12 g lysine/MJ DE). The selected lines resulted from seven generations of selection for high daily food intake, lean food conversion ratio (LFC) and lean growth rate on ad libitum (LGA) or restricted (LGS) feeding regimes in a Large White population. During performance test, from 30 to 90 kg, boars were fed on either ad libitum or restricted (0.75 g/g ad libitum daily food intake) feeding regimes. A sensory panel assessed heated fat samples for androstenone odour, skatole odour and abnormal odour. There were no signi®cant dierences between the selection and control lines or diets for log transformed fat content of androstenone, skatole, indole. The signi®cant diet with feeding regime interaction for log transformed fat content of skatole and indole were essentially due to signi®cantly higher log transformed fat contents with ad libitum feeding of the high lysine diet compared to restricted feeding (skatole: ÿ1.94 vs ÿ3.06, s.e.d. 0.43; indole: ÿ3.44 vs ÿ4.22, s.e.d. 0.28), as dierences between feeding regimes on diets A and C were not signi®cantly dierent from zero. There were no signi®cant dierences between selection and control lines for sensory panel score for abnormal odour or androstenone odour, but the LFC and LGA selection lines had a signi®cantly higher skatole odour score than the LGS selection line. Neither diet nor feeding regime had any signi®cant eect on sensory panel assessment of odour. Log transformed fat content of androstenone and skatole were signi®cantly correlated with sensory panel score for skatole odour (0.37 and 0.46, s.e. 0.12), but not with sensory panel score for androstenone odour (0.06 and 0.09), such that they would not be useful predictors of androstenone odour. # 1999 Elsevier Science Ltd. All rights reserved.
1. Introduction Boar taint is an objectionable odour detected during cooking of meat, with androstenone (5a-androst-16-en3-one) and skatole (3-methyl-indole) thought to contribute to boar taint (Bonneau, 1997). Androstenone is synthesised in the testes of sexually mature boars and released into the blood, while skatole is formed in the colon resorbed into the blood and both accumulate in adipose tissue (Claus, Weiler & Herzog, 1994). Betweenbreed variation in fat content of androstenone and skatole and in sensory panel assessment of boar taint has
* Corresponding author. Tel.: +44-131-527-4200; fax: +44-131440-0434.
been demonstrated (Xue et al., 1996), as has withinbreed variation in fat content of androstenone (Willeke, 1993). If selection for particular components of ecient lean growth rate changed the shape of the growth curve or reduced mature size, such that animals reached sexual maturity earlier, then there may be a correlated response in fat content of androstenone and skatole at a given slaughter weight, resulting in an increase in boar taint. There is evidence for a major gene for fat androstenone content (Fouilloux, Le Roy, Gruand, Renard, Sellier & Bonneau, 1997) and a suggestion of a recessive gene for fat skatole content (LundstroÈm et al., 1994). Therefore, particular selection strategies may change the frequency of a favourable allele, if the gene is linked to genes controlling traits in the selection criterion. Evaluation of alternative selection strategies should
0309-1740/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved. PII: S0309-1740(99)00088-1
148
N.D. Cameron et al. / Meat Science 54 (2000) 147±153
incorporate examination of the responses in boar taint and its contributing factors. Further, if diets with low lysine content can reduce the fat content of skatole (Andersson, Schaub, Andersson, LundstroÈm, Thomke & Hansson, 1997), such that deleterious responses in boar taint may be countered by dietary treatment, then it would be necessary to determine if a genotype with nutrition interaction exists for boar taint and its component traits. Study of genotype with nutrition interactions for performance test traits, carcass composition and meat quality, examined using the selection lines from the Edinburgh lean growth selection experiment (Cameron, Penman, Fisken, Nute, Perry & Wood, 1999c) was extended to include boar taint and fat content of androstenone, skatole and indole in boars.
Table 1 Composition and speci®cation of diets (g/kg) Diet A
C
E
Wheat meal Barley meal Soyabean meal Fish meal Maize gluten meal Maize oil Lysine Hcl Vitamins and minerals Dicalcium phosphate Limestone ¯our Sodium chloride
600 200 150 0 15 17 0 3 5 8 2
340 339 210 40 40 11 2.5 3 5 7.5 2
80 478 270 80 64 5 5 3 5 8 2
DE MJ/kg Crude protein (g/kg) Ileal lysine (g/kg)
14.0 157 5.64
14.0 210 10.68
14.0 262 15.72
2. Materials and methods 2.1. Animals and performance test Details on establishment of the Large White population and the four selection groups, with divergent selection over seven generations for daily food intake (DFI), lean food conversion ratio (LFC) and lean growth rate on ad libitum (LGA) or restricted (LGS) feeding regimes, were given by Cameron (1994). The objectives of the LFC, LGA and LGS selection groups were to achieve equal correlated responses, measured in phenotypic s.d., for carcass lean content and food conversion ratio (LFC) or growth rate (LGA and LGS). A control line was also established for comparison with the selection lines. The selection lines were performance tested in separate batches, due to the batch-farrowing management system, but boars in the control line were tested in each batch, such that environmental dierences between batches could be accounted for in the analyses. Animals from the four high selection lines and the control lines were tested on ad libitum or restricted (0.75 g/g ad libitum daily food intake) feeding regimes with three isoenergetic diets (14.0 MJ DE/kg DM) diering in ileal digestible lysine to digestible energy (DE) (A: 0.40, C: 0.76 and E: 1.12 g lysine/MJ DE). The coding of the three diets is consistent with the ®ve diets used in the corresponding genotype with nutrition interaction study for performance test traits (Cameron & MacLeod, 1997). The objective of current study was to determine if a genotype with nutrition interaction existed, such that the two extreme diets (A and E) and a ``standard'' test diet C were chosen for the study. Van Lunen and Cole (1996) reported an optimum lysine: MJ DE ratio of 0.95±1 and the lysine: MJ DE ratio of diet C was 0.97. The composition and speci®cation of the diets are given in Table 1. The diets were required to meet three lysine: digestible energy ratios with a constant DE and that the
ingredients were broadly the same for each diet. Inclusion of ®sh meal in the diets was unlikely to have implications on palatability (Valaja, Suomi, Alaviuhkola & Immonen, 1992). Equal protein: lysine ratios for the three diets could not be achieved using the set of dietary ingredients, but the diets were formulated to ensure that lysine was always the ®rst-limiting amino acid (M. MacLeod, pers. comm.). The performance test was on a ®xed weight basis, with start and ®nish weights of 30 (3) kg and 90 (5) kg, and individual penning of all animals. There were 64 animals in the study. In each of the four selection lines 12 boars were tested on the three diets and two feeding regimes with two animals per dietfeeding regime subclass. A total of 16 boars from the control line were tested on the two feeding regimes, but only on diet C. Food allocation of animals on the restricted feeding regime was 1.1 kg/day at the start of test and was incremented by 0.1 kg/day each week of test, up to a maximum of 2.3 kg/day. After completion of performance test, animals were transported from Edinburgh to Bristol. At Bristol, animals were fed the same diet and ration as at the end of the performance test, until slaughter 5 days later. Pigs were then moved to the slaughter-house and lairaged overnight to allow them to settle. Food was withdrawn no more than 12 h before slaughter and water was freely available. Muscle and eating quality traits were measured in the loin joints from the right hand side of each carcass (Cameron, Nute, Brown, Enser & Wood, 1999b). The longissimus dorsi muscle maximum depth (measured on the cut surface at the last rib) and the depth of subcutaneous fat, above the maximum muscle depth, were measured on the right-hand side of the cold carcass, but only in the DFI, LFC and LGA lines and the control line. Each animal was assessed for abnormal
N.D. Cameron et al. / Meat Science 54 (2000) 147±153
odour of cooked meat (1=low and 8=high intensity) by a sensory panel (Wood, Nute, Fursey, & Cuthbertson, 1995). Carcass subcutaneous fat weight was predicted from data on the dissected fore-loin joint, using a prediction equation determined from carcass data in a previous generation (Cameron & Curran, 1995): Carcass subcutaneous fat weight 3
Cwt Jwt i Jwti i1
where Cwt and Jwt are the carcass and fore-loin joint weights and Jwti are the weights of lean, subcutaneous fat and bone in the fore loin joint. The proportion of variation in carcass subcutaneous fat weight accounted for by the regression equation was 0.87. 2.2. Androstenone, skatole and indole measurement and sensory assessment of fat Fat samples were obtained from over the longissimus dorsi muscle. Androstenone concentration in fat was measured using the procedure described by De Brabander, Verbeke, Dirinck and Casteels (1985), with skatole and indole concentrations determined by the Likens-Nickerson method (Annor-Frempong, Nute, Whittingdon & Wood, 1997a). For sensory analysis, the fat samples were prepared as described by AnnorFrempong, Nute, Whittingdon and Wood (1997b). Brie¯y, 11±13 g samples of fat, taken from over the longissimus dorsi muscle, were individually heated for up to 8 min in Tecam heating blocks set at 100 C to a maximum of 65 C and then placed in 120 ml amber glass screw-top bottles. The sensory panel consisted of 10 females, aged 30±60 years, and each panellist was situated in an individual, well-ventilated booth. The panellists were selected for their ability to repeatedly distinguish between samples diering in fat content of androstenone and skatole. Four fat samples were given to each panelist in each session, with four sessions per day and 30 min between sessions. Each sample was assessed for intensity of abnormal odour, androstenone odour and skatole odour. The odour intensities were assessed on eight point scales (1=low and 8=high intensity). 2.3. Statistical analysis The genotype with nutrition interaction was estimated by residual maximum likelihood (REML) analyses using the REML algorithm of Genstat 5.3 Committee (1993). The model included ®xed eects of selection line, diet, feeding regime and two-way interactions between
149
selection line, diet and feeding regime, with batch included as a random eect. The three-way interaction of selection line, diet and feeding regime was initially included in the model to determine if the selection line with diet interaction was feeding regime dependent. For all traits, the three-way interaction was not statistically signi®cant and was deleted from the model. Cold carcass weight was included in the model as a covariate, as the experiment was designed to measure animals on a ®xed weight basis. Log transformations of fat content of androstenone, skatole and indole were required, as the traits were positively skewed. Throughout the text, all eects which are dierent at the 0.05 signi®cance level are referred to as being signi®cant. The probabilities of having a fat androstenone or skatole concentration above the thresholds of 1 or 0.25 mg/g, respectively, (Bonneau, 1997) were determined by analysing the binomial traits for each animal, denoting concentrations below or above the thresholds. The binomial traits were analysed using a generalised linear mixed model (Genstat 5.3 Committee, 1993). The model had a binomial error and a logit link (Schall, 1991; Welham, 1993) and included selection line, diet and feeding regime as ®xed eects, the two-way interactions between ®xed eects and batch as a random eect. Sensory panel scores for abnormal odour, androstenone odour and skatole odours were not transformed for analysis, as the precision of estimated betweenselection line dierences was not increased after transformation of sensory panel scores to standardised normal deviates (Cameron et al., 1999b). 3. Results 3.1. Growth and carcass traits Dierences between the selection and control lines, dietary and feeding regime eects on carcass traits are presented in Table 2. Between-selection line variation in cold carcass weight resulted from dierences in weight of transportation to Bristol, as when that weight was included in the model as a covariate for cold carcass weight, there were no dierences between the selection lines. The control line was intermediate to the LFC and DFI selection lines for both carcass content and growth rate of subcutaneous fat. The LGA and LGS lines had similar rates of subcutaneous fat growth as the control line. Diet A resulted in signi®cantly greater subcutaneous fat depth and higher predicted carcass content and growth rate of fat than diets C and E, which were similar. Predicted carcass content and growth rate of fat were also signi®cantly higher for ad libitum than restricted fed boars. There were no statistically signi®cant genotype with nutrition interactions for carcass traits.
150
N.D. Cameron et al. / Meat Science 54 (2000) 147±153
Table 2 Mean values for boar taint traits of the four selection lines (lean growth rate with ad-libitum, LGA or restricted, LGS feeding; lean food conversion, LFC; daily food intake, DFI), relative to the control line, of the three diets and two feeding regimes Genotype
Diet
Feeding regime
Control DFIa
LFC
LGA
LGS
s.e.d.
A
C
E
s.e.d.
ad libb restr
s.e.d.
Cold carcass weight (kg) Subcutaneous fat depth (mm) Predicted carcass sub. fat (g/kg) Pred. sub. fat growth rate (g/day)
67.9 12.1 255 212
7.7 ÿ0.8 46 58
0.5 ÿ4.4 ÿ53 ÿ49
ÿ5.6 ÿ5.2 ÿ34 9
0.3 ± ÿ30 ÿ15
3.6 1.7 18 17
69.2 12.5 282 246
68.0 8.2 220 204
68.3 7.7 220 189
1.9 1.0 10 15
69.6 9.4 254 244
67.4 9.6 228 183
1.3 0.8 7.4 10
Log androstenone Log skatole Log indole
0.15 ÿ2.29 ÿ3.73
ÿ0.18 0.11 0.17
ÿ0.05 ÿ0.55 ÿ0.01
0.05 0.09 ÿ0.03
ÿ0.32 ÿ0.53 0.06
0.23 0.45 0.26
0.12 ÿ2.42 ÿ3.56
0.03 ÿ2.45 ÿ3.66
0.01 ÿ2.52 ÿ3.85
0.15 0.32 0.20
0.21 ÿ2.20 ÿ3.52
ÿ0.11 ÿ2.73 ÿ3.86
0.11 0.23 0.15
0.51 0.0
0.05 0.2
0.10 0.4
ÿ0.23 ÿ1.0
ÿ0.24 ÿ1.0
1.4
0.44 ÿ0.3
0.54 0.2
0.34 ÿ0.7
0.8
0.62 0.5
0.27 ÿ1.0
0.6
0.09 ÿ2.3
0.08 0.8
0.02 0.2
0.17 1.3
ÿ0.08 ÿ2.3
1.4
0.10 ÿ2.2
0.13 ÿ1.9
0.11 ÿ2.1
0.07 ÿ2.6
0.7
4.42 2.89 5.35 3.04
ÿ0.25 0.18 ÿ0.03 0.06
ÿ0.43 0.36 ÿ0.21 0.17
ÿ0.28 0.56 ÿ0.25 0.42
ÿ0.04 ÿ0.38 ÿ0.02 ÿ0.11
0.35 0.33 0.37 0.30
4.14 3.00 5.15 3.19
4.21 3.02 5.12 3.14
4.21 3.02 5.25 3.12
4.24 3.04 5.24 3.18
P(androstenone>1 mg/g) Logit score=log[1/(1ÿprob)] P(skatole>0.25 mg/g) Logit score Fat: androstenone odour (1:weak) Fat: skatole odour (1:weak) Fat: abnormal odour (1:weak) Meat: abnormal odour (1:weak) a b
0.05 ÿ3.0 10 4.32 3.07 5.47 3.11
0.17 0.23 0.20 0.25
0.12 0.17 0.14 0.18
Dierence between the DFI, LFC, LGA and LGS genotypes and the control line. ad libitum and restricted feeding regimes.
3.2. Androstenone, skatole and indole Average fat contents of androstenone, skatole and indole were 1.05, 0.085 and 0.025 mg/g, respectively, but as log transformations were required for the statistical analyses, mean values of log transformed fat content of androstenone, skatole, indole are given in Table 2. There were no signi®cant dierences between the selection and control lines or diets for log transformed fat content of androstenone, skatole, indole. However, the probability of a skatole concentration of fat greater than 0.25 mg/g was signi®cantly higher in the DFI and LGA selection lines than in the LGS line. Ad libitum fed boars had signi®cantly higher log transformed fat content of androstenone, skatole and indole than restricted fed boars and the probability of an androstenone concentration greater than 1 was signi®cantly higher. There were signi®cant diet with feeding regime interactions for log transformed fat contents of skatole and indole, essentially due to signi®cantly higher log transformed fat contents with ad libitum feeding of diet E compared to restricted feeding (skatole: ÿ1.94 vs ÿ3.06, s.e.d. 0.43; indole: ÿ3.44 vs ÿ4.22, s.e.d. 0.28). Ad libitum feeding of diets A and C resulted in higher log transformed fat contents of androstenone, skatole and indole, but the within-diet contrasts were not signi®cantly dierent from zero. 3.3. Sensory assessment of fat There were no signi®cant dierences between selection and control lines for sensory panel score for
androstenone odour, but the LFC and LGA selection lines had a signi®cantly higher sensory panel score for skatole odour than the LGS selection line (Table 2). Both diet and feeding regime had no signi®cant eect on sensory panel assessment. There was a signi®cant selection line with feeding regime interaction for skatole odour, with a higher sensory panel score for ad libitum fed than for restricted fed LGA boars (3.87 vs 2.77, s.e.d. 0.32). A statistically signi®cant selection line with diet interaction was detected for sensory panel score of androstenone odour, due to higher scores for the LGA selection line on diet C compared to diets A and E (4.40 vs 3.71 and 3.81, s.e.d. 0.32), with the converse for the LGS line (4.00 vs 4.47 and 4.85). 3.4. Correlations between traits Correlations between traits are presented in Table 3. Log transformed fat content of androstenone, skatole and indole were signi®cantly positively correlated, particularly androstenone and skatole (Fig. 1). Log transformed fat contents of androstenone and skatole were signi®cantly correlated with sensory panel score for skatole odour, but not with panel score for androstenone odour. Sensory panel scores for androstenone odour and skatole odour were both signi®cantly correlated with sensory panel score for abnormal odour, although the former two traits were essentially independent. Animals were classed as low androstenone (LA)-low skatole (LS), high androstenone (HA)-LS, and HA-high skatole using the thresholds of 1 and 0.25 mg/g for fat content of androstenone and skatole,
N.D. Cameron et al. / Meat Science 54 (2000) 147±153
151
Table 3 Correlations (100) between fat content of androstenone, skatole, indole and sensory panel assessmenta
Log androstenone (AND) Log skatole (SKA) Log indole (IND) Fat: Androstenone odour (O-AND) Fat: Skatole odour (O-SKA) Fat: abnormal odour (O-ABN) Meat: abnormal odour (M-ABN) Predicted carcass sub. fat (FAT) Predicted sub. fat growth rate a
AND
SKA
IND
O-AND
O-SKA
O-ABN
M-ABN
FAT
80a 36 6 37 29 40 41 48
41 9 46 46 60 53 52
18 10 34 23 28 33
14 62 10 6 7
56 32 0 2
40 30 28
40 46
90
s.e. of correlations =0.12.
for sensory panel scores for skatole odour (1.36, s.e. 0.46 and 0.20, s.e. 0.10) and abnormal odour (1.13, s.e. 0.38 and 0.17, s.e. 0.08). The linear and quadratic regression coecients of log transformed fat content of indole were signi®cantly negative (ÿ3.8, s.e. 1.8 and ÿ0.50, s.e. 0.24) as explanatory variables for sensory panel scores for skatole odour. The coecients of determination (R2) of the models for sensory panel scores for skatole and abnormal odours were only 0.30 and 0.33, respectively, with no change in the latter when log transformed fat contents of androstenone or indole or both were dropped from the model. None of the regression coecients were signi®cantly dierent from zero with sensory panel score for androstenone odour as the dependent variable, in line with the corresponding correlation coecients. Fig. 1. Log transformed fat contents of androstenone and skatole.
respectively. However, the relationship between fat content and sensory panel odour score for androstenone and skatole with animals classed as LA-LS, LA-HS, HA-LS, and HA-HS (Annor-Frempong et al., 1997b) could not be examined, as there were insucient numbers of observations in the LA-HS and HA-LS classes (see Fig. 1). Predicted carcass content and growth rate of fat were signi®cantly correlated with log transformed fat content of androstenone and skatole, but not with sensory panel score for androstenone odour or skatole odour. Relationships between sensory panel androstenone, skatole and abnormal odour scores with linear and quadratic log transformed androstenone, skatole and indole fat content were examined in multiple regression analyses. Regression coecients for log transformed fat content of androstenone were not signi®cantly dierent from zero for sensory panel androstenone, skatole or abnormal odour scores. Linear and quadratic terms of log transformed skatole fat content were statistically signi®cantly dierent from zero as explanatory variables
4. Discussion The study was performed prior to the availability of results from associated meat quality studies using animals from the lean growth experiment. In retrospect, the general lack of signi®cant dierences in fat content of androstenone or in sensory panel assessment of boar taint traits in fat between the selection lines or diets for boars slaughtered at 90 kg was not unexpected. Firstly, there were no signi®cant dierences in eating quality between the selection lines, when performance tested on a single diet (Cameron et al., 1999b) or on the same three diets as in the current study (Cameron et al., 1999c). Secondly, it was unlikely that boars in the current study had reached sexual maturity as the average weight at ®rst oestrus of gilts from the selection lines was 124 kg, with ®rst oestrus determined from weekly plasma progesterone assay, (Cameron, Kerr, Garth & Sloan, 1999a). However, the study did detect a signi®cantly lower sensory panel score for skatole odour of fat in the LGS line compared to the LFC and LGA lines, such that the probability of a greater fat content of skatole than the threshold of 0.25 mg/g was lower in
152
N.D. Cameron et al. / Meat Science 54 (2000) 147±153
the LGS line. Therefore, selection for lean growth rate with performance testing on a restricted feeding regime, to produce animals of high genetic merit for ecient lean growth rate, may be associated with a lower incidence of boar taint, with regard to fat content of skatole. LundstroÈm et al. (1994) reported that selection for high lean growth on a high protein diet (185 g/kg) resulted in a lower fat content of skatole than selection on a low protein diet (131 g/kg). In the current study, there was no eect of diet on fat content of skatole given the range of dietary protein contents of 157, 210 and 262 g/kg, which was substantially greater than in the studies of LundstroÈm et al. (1988) (144 and 163 g/ kg), LundstroÈm et al. (1994), Andersson et al. (1997) (138 and 141 g/kg) and Nold, Romans, Costello, Henson and Libal (1997) (120±170 g/kg), which also reported no consistent signi®cant eect of diet. Therefore, dierences in fat content of skatole in the selection lines of LundstroÈm et al. (1994) may not have been due to dierences in dietary protein content - a point also made by LundstroÈm et al. (1994). Further, the selection lines of LundstroÈm et al. (1994) diered to the same extent in fat content of skatole in generations one, two and four, suggesting that the higher frequency of the allele for high fat content of skatole in the line selected on a low protein diet was due to serendipity, given the assumption of LundstroÈm et al. (1994) that a single gene exists for fat content of skatole. The diet with feeding regime interaction for fat content of skatole and indole, due to signi®cantly higher fat contents with ad libitum feeding than with restricted feeding for diet E than for diets A and C, may primarily be a consequence of higher intake of dietary tryptophan, as skatole and indole are formed from intestinal degradation of tryptophan (Claus et al., 1994). Diet E was included in the study to generate a sucient range of diets, such that dietary eects could be detected. If diet E had not been included in the study, then the diet with feeding regime interaction on log transformed fat contents of androstenone, skatole or indole would not have been detected. In previous studies, the range of diets has been too small that it was unlikely that a dietary eect could have been detected, if such an eect existed. Therefore, it is important that nutritional studies include diets which dier suciently to generate the necessary experimental power to detect dietary dierences. Despite substantial dierences in carcass fat content and fat growth rate between the DFI, control and LFC selection lines, there were no signi®cant dierences in log transformed fat content of androstenone, skatole and indole. Similarly, there were signi®cant dierences in carcass fat content and fat growth rate between diets, but they were not accompanied by dierences in log transformed fat content of androstenone, skatole and indole. Therefore, neither the rate of fat deposition nor
the carcass content of fat per se contribute consistently to the fat content of androstenone, skatole and indole, despite statistically signi®cant positive correlations between carcass content and growth rate of fat with log transformed fat content of androstenone, skatole and indole. Log transformed fat content of androstenone and indole provided no additional information on sensory panel score for abnormal odour in the current study, that was not provided by log transformed fat content of skatole. In contrast, Annor-Frempong et al. (1997b) reported that linear and quadratic terms for fat content of androstenone, skatole and indole proportionally accounted for 0.76 of the variation in sensory panel score for abnormal odour. The diering conclusions in the two studies may have stemmed from dierences in the observed variation in fat content of androstenone and in the statistical analyses. The minimum fat content of androstenone in the current study was 0.37 mg/g, while gilts were included in the Annor-Frempong et al. (1997b) study, such that half the animals had a fat content of androstenone of less than 0.30 mg/g. Despite the skewed distributions for fat content of androstenone, skatole and indole, the data were not transformed in the Annor-Frempong et al. (1997b) study, such that standard errors of the regression coecients may have been under-estimated resulting in false positive conclusions. Studies designed to determine the predictive value of fat content of androstenone, skatole and indole for sensory panel assessment of abnormal odour should dierentiate between boars and gilts and should account for the non-normal distributions of fat content of androstenone, skatole and indole. Based on the numbers of animals in the study, the power of the experiment appears to be limited for detection of statistically signi®cant dierences in boar taint traits between the selection lines and the control line. The experimental power was determined using the formula of Hill (1978) and assumptions regarding the estimation of drift and sampling variances are given by Cameron, et al. (1999c). To achieve an experimental power of 0.80 for detection of signi®cant correlated responses in boar taint traits relative to the control line, the parameter rA h needed to exceed 0.18, where rA is the genetic correlation of a boar taint trait with the selection criterion and h2 is the heritability of the boar taint trait. For example, a genetic correlation of 0.4 and a heritability of at least 0.2 would be sucient, which was thought reasonable when the experiment was designed. Even if the experiment was quadrupled in size, the required minimum value of rA h would only reduce to 0.16, such that a genetic correlation of 0.3 would suce. The experimental power was enhanced by measuring the response in selection lines relative to the control line, when the lines were derived from the same base population, rather than from having a high number of
N.D. Cameron et al. / Meat Science 54 (2000) 147±153
animals per selection line. However, it would be instructive to compare standard errors of dierences in the current study with those in previous studies, but literature information on the variation of studied traits is limited, as several recent studies (Bonneau, le Denmat, Vaudelet, Veloso Nunes, Mortensen & Mortensen, 1992; Xue et al., 1996; Annor-Frempong et al., 1997b) have not reported either standard deviations or standard errors of dierences between ``treatments''. In conclusion, there was no evidence of selection strategy speci®c responses in fat content of androstenone, skatole or indole, traits associated with boar taint. However, selection for rate of lean growth on a restricted feeding regime resulted in a lower sensory panel score for skatole odour than selection for either rate or eciency on lean growth on ad libitum feeding. The higher fat content of skatole and indole with ad libitum feeding compared to restricted feeding of a diet with a high ileal digestible lysine content may have primarily been a consequence of higher intake of dietary tryptophan. While fat content of androstenone and skatole could be used as predictors of sensory panel odour score for skatole, they would not be useful predictors of androstenone odour. Acknowledgements The project was funded by the Ministry of Agriculture, Fisheries and Food. The diets were designed by M. MacLeod, Roslin Institute. References Andersson, K., Schaub, A., Andersson, K., LundstroÈm, K., Thomke, S., & Hansson, I. (1997). The eects of feeding system, lysine level and gilt contact on performance, skatole levels and economy of entire male pigs. Livestock Production Science, 51, 131±140. Annor-Frempong, I. E., Nute, G. R., Whittington, F. W., & Wood, J. D. (1997a). The problem of taint in pork II. The in¯uence of skatole, androstenone and indole, presented individually and in combination in a model lipid base, on odour perception. Meat Science, 47, 49±61. Annor-Frempong, I. E., Nute, G. R., Whittington, F. W., & Wood, J. D. (1997b). The problem of boar taint in pork III. Odour pro®le of pork fat and the interrelationships between androstenone, skatole and indole concentrations. Meat Science, 47, 63±76. Bonneau, M., le Denmat, M., Vaudelet, J. C., Veloso Nunes, J. R., Mortensen, A. B., & Mortensen, H. P. (1992). Contributions of fat androstenone and skatole to boar taint: I. Sensory attributes of fat and pork meat. Livestock Production Science, 32, 63±80. Bonneau, M. (1997). Boar taint in entire male pigs Ð a review. Pig News and Information, 18, 15N±18N. Cameron, N. D. (1994). Selection for components of ecient lean growth rate in pigs. 1. Selection pressure applied and direct responses in a Large White herd. Animal Production, 59, 251±262.
153
Cameron, N. D., & Curran, M. K. (1995). Responses in carcass composition to divergent selection for components of ecient lean growth rate in pigs. Animal Science, 61, 347±359. Cameron, N. D., & MacLeod, M. G. (1997). Genotype with nutrition interaction for performance test traits in pigs selected for lean growth rate. Proceedings of the British Society of Animal Science Annual Meeting. Paper 29. Cameron, N. D., Kerr, J. C., Garth, G. B., & Sloan, R. L. (1999a). Genetic and nutritional eects on age at ®rst oestrus of gilts selected for components of ecient lean growth rate. Animal Science, 69, 93±103. Cameron, N. D., Nute, G. R., Brown, S. N., Enser, M., & Wood, J. D. (1999b). Meat quality of Large White pig genotypes selected for components of ecient lean growth rate. Animal Science, 68, 115±127. Cameron, N. D., Penman, J. C., Fisken, A. C., Nute, G. R., Perry, A. M., & Wood, J. D. (1999c). Genotype with nutrition interactions for carcass composition and meat quality in pig genotypes selected for components of ecient lean growth rate. Animal Science, 69, 69±80. Claus, R., Weiler, U., & Herzog, A. (1994). Physiological aspects of androstenone and skatole formation in the boar Ð a review with experimental data. Meat Science, 38, 289±305. De Brabander, H. E., Verbeke, R., Dirinck, P., & Casteels, M. (1985). Boar taint in Belgian pigs in relation to the androstenone content. Proc. 31st Europ. Meet. Res. Workers. Varna, 778±781. Fouilloux, M. N., Le Roy, P., Gruand, J., Renard, C., Sellier, P., & Bonneau, M. (1997). Support for single major genes in¯uencing fat androstenone level and development of bulbo-urethral glands in young boars. Genetics Selection Evolution, 29, 357±366. Genstat 5.3 Committee (1993). Genstat 5.3 reference manual. Oxford: Clarendon Press. Hill, W. G. (1978). Design of selection experiments for comparing alternative testing regimes. Heredity, 41, 371±376. LundstroÈm, K., Malmfors, B., Malmfors, G., Stern, S., Petersson, H., Mortensen, A. B., & Sùrensen, S. E. (1988). Skatole, androstenone and taint in boars fed two dierent diets. Livestock Production Science, 18, 55±67. LundstroÈm, K., Malmfors, B., Stern, S., Rydhmer, L., Eliasson-Selling, L., Mortensen, A. B., & Mortensen, H. P. (1994). Skatole levels in pigs selected for high lean tissue growth rate on dierent dietary protein levels. Livestock Production Science, 38, 125±132. Nold, R. A., Romans, J. R., Costello, W. J., Henson, J. A., & Libal, G. W. (1997). Sensory characteristics and carcass traits of boars, barrows and gilts fed high- or adequate-protein diets and slaughtered at 100 or 110 kilograms. Journal of Animal Science, 75, 2641± 2651. Schall, R. (1991). Estimation of generalised linear models with random eects. Biometrika, 78, 719±727. Valaja, J., Suomi, K., Alaviuhkola, T., & Immonen, I. (1992). Eect of dietary ®sh-meal on the palatability and fatty-acid composition of pork. Agricultural Science in Finland, 1, 21±26. Van Lunen, T. A., & Cole, D. J. A. (1996). The eect of lysine/digestible energy ratio on growth performance and nitrogen deposition of hybrid boars, gilts and castrated male pigs. Animal Science, 63, 465± 475. Welham, S. J. (1993). The GLMM procedure. Genstat 5, Procedure library manual release, 3(1), 187±192. Willeke, H. (1993). Possibilities of breeding for low 5a-androstenone content in pigs. Pig News and Information, 14, 31N±33N. Wood, J. D., Nute, G. R., Fursey, G. A. J., & Cuthbertson, A. (1995). The eect of cooking conditions on the eating quality of pork. Meat Science, 40, 127±135. Xue, J. L., Dial, G. D., Holton, E. E., Vickers, Z., Squires, E. J., Loy, Y. P., Godbout, D., & Morel, N. (1996). Breed dierences in boar taint: relationships between tissue levels of boar taint compounds and sensory analysis of taint. Journal of Animal Science, 74, 2170±2177.
www.elsevier.com/locate/meatsci
Boar taint in pigs selected for components of ecient lean growth rate N.D. Cameron a,*, J.C. Penman a, A.C. Fisken a, G.R. Nute b, A.M. Perry b, F.W. Whittington b a Roslin Institute (Edinburgh), Roslin, Midlothian EH25 9PS, UK Division of Food Animal Science, School of Veterinary Science, University of Bristol, Langford BS4 05U, UK
b
Received 29 March 1999; accepted 3 July 1999
Abstract Genetic and nutritional eects on the boar taint traits of androstenone, skatole and indole and the eating quality traits associated with boar taint were examined by testing animals from four selection lines and a control line on isoenergetic diets, which diered in ileal digestible lysine: digestible energy (0.40, 0.76 and 1.12 g lysine/MJ DE). The selected lines resulted from seven generations of selection for high daily food intake, lean food conversion ratio (LFC) and lean growth rate on ad libitum (LGA) or restricted (LGS) feeding regimes in a Large White population. During performance test, from 30 to 90 kg, boars were fed on either ad libitum or restricted (0.75 g/g ad libitum daily food intake) feeding regimes. A sensory panel assessed heated fat samples for androstenone odour, skatole odour and abnormal odour. There were no signi®cant dierences between the selection and control lines or diets for log transformed fat content of androstenone, skatole, indole. The signi®cant diet with feeding regime interaction for log transformed fat content of skatole and indole were essentially due to signi®cantly higher log transformed fat contents with ad libitum feeding of the high lysine diet compared to restricted feeding (skatole: ÿ1.94 vs ÿ3.06, s.e.d. 0.43; indole: ÿ3.44 vs ÿ4.22, s.e.d. 0.28), as dierences between feeding regimes on diets A and C were not signi®cantly dierent from zero. There were no signi®cant dierences between selection and control lines for sensory panel score for abnormal odour or androstenone odour, but the LFC and LGA selection lines had a signi®cantly higher skatole odour score than the LGS selection line. Neither diet nor feeding regime had any signi®cant eect on sensory panel assessment of odour. Log transformed fat content of androstenone and skatole were signi®cantly correlated with sensory panel score for skatole odour (0.37 and 0.46, s.e. 0.12), but not with sensory panel score for androstenone odour (0.06 and 0.09), such that they would not be useful predictors of androstenone odour. # 1999 Elsevier Science Ltd. All rights reserved.
1. Introduction Boar taint is an objectionable odour detected during cooking of meat, with androstenone (5a-androst-16-en3-one) and skatole (3-methyl-indole) thought to contribute to boar taint (Bonneau, 1997). Androstenone is synthesised in the testes of sexually mature boars and released into the blood, while skatole is formed in the colon resorbed into the blood and both accumulate in adipose tissue (Claus, Weiler & Herzog, 1994). Betweenbreed variation in fat content of androstenone and skatole and in sensory panel assessment of boar taint has
* Corresponding author. Tel.: +44-131-527-4200; fax: +44-131440-0434.
been demonstrated (Xue et al., 1996), as has withinbreed variation in fat content of androstenone (Willeke, 1993). If selection for particular components of ecient lean growth rate changed the shape of the growth curve or reduced mature size, such that animals reached sexual maturity earlier, then there may be a correlated response in fat content of androstenone and skatole at a given slaughter weight, resulting in an increase in boar taint. There is evidence for a major gene for fat androstenone content (Fouilloux, Le Roy, Gruand, Renard, Sellier & Bonneau, 1997) and a suggestion of a recessive gene for fat skatole content (LundstroÈm et al., 1994). Therefore, particular selection strategies may change the frequency of a favourable allele, if the gene is linked to genes controlling traits in the selection criterion. Evaluation of alternative selection strategies should
0309-1740/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved. PII: S0309-1740(99)00088-1
148
N.D. Cameron et al. / Meat Science 54 (2000) 147±153
incorporate examination of the responses in boar taint and its contributing factors. Further, if diets with low lysine content can reduce the fat content of skatole (Andersson, Schaub, Andersson, LundstroÈm, Thomke & Hansson, 1997), such that deleterious responses in boar taint may be countered by dietary treatment, then it would be necessary to determine if a genotype with nutrition interaction exists for boar taint and its component traits. Study of genotype with nutrition interactions for performance test traits, carcass composition and meat quality, examined using the selection lines from the Edinburgh lean growth selection experiment (Cameron, Penman, Fisken, Nute, Perry & Wood, 1999c) was extended to include boar taint and fat content of androstenone, skatole and indole in boars.
Table 1 Composition and speci®cation of diets (g/kg) Diet A
C
E
Wheat meal Barley meal Soyabean meal Fish meal Maize gluten meal Maize oil Lysine Hcl Vitamins and minerals Dicalcium phosphate Limestone ¯our Sodium chloride
600 200 150 0 15 17 0 3 5 8 2
340 339 210 40 40 11 2.5 3 5 7.5 2
80 478 270 80 64 5 5 3 5 8 2
DE MJ/kg Crude protein (g/kg) Ileal lysine (g/kg)
14.0 157 5.64
14.0 210 10.68
14.0 262 15.72
2. Materials and methods 2.1. Animals and performance test Details on establishment of the Large White population and the four selection groups, with divergent selection over seven generations for daily food intake (DFI), lean food conversion ratio (LFC) and lean growth rate on ad libitum (LGA) or restricted (LGS) feeding regimes, were given by Cameron (1994). The objectives of the LFC, LGA and LGS selection groups were to achieve equal correlated responses, measured in phenotypic s.d., for carcass lean content and food conversion ratio (LFC) or growth rate (LGA and LGS). A control line was also established for comparison with the selection lines. The selection lines were performance tested in separate batches, due to the batch-farrowing management system, but boars in the control line were tested in each batch, such that environmental dierences between batches could be accounted for in the analyses. Animals from the four high selection lines and the control lines were tested on ad libitum or restricted (0.75 g/g ad libitum daily food intake) feeding regimes with three isoenergetic diets (14.0 MJ DE/kg DM) diering in ileal digestible lysine to digestible energy (DE) (A: 0.40, C: 0.76 and E: 1.12 g lysine/MJ DE). The coding of the three diets is consistent with the ®ve diets used in the corresponding genotype with nutrition interaction study for performance test traits (Cameron & MacLeod, 1997). The objective of current study was to determine if a genotype with nutrition interaction existed, such that the two extreme diets (A and E) and a ``standard'' test diet C were chosen for the study. Van Lunen and Cole (1996) reported an optimum lysine: MJ DE ratio of 0.95±1 and the lysine: MJ DE ratio of diet C was 0.97. The composition and speci®cation of the diets are given in Table 1. The diets were required to meet three lysine: digestible energy ratios with a constant DE and that the
ingredients were broadly the same for each diet. Inclusion of ®sh meal in the diets was unlikely to have implications on palatability (Valaja, Suomi, Alaviuhkola & Immonen, 1992). Equal protein: lysine ratios for the three diets could not be achieved using the set of dietary ingredients, but the diets were formulated to ensure that lysine was always the ®rst-limiting amino acid (M. MacLeod, pers. comm.). The performance test was on a ®xed weight basis, with start and ®nish weights of 30 (3) kg and 90 (5) kg, and individual penning of all animals. There were 64 animals in the study. In each of the four selection lines 12 boars were tested on the three diets and two feeding regimes with two animals per dietfeeding regime subclass. A total of 16 boars from the control line were tested on the two feeding regimes, but only on diet C. Food allocation of animals on the restricted feeding regime was 1.1 kg/day at the start of test and was incremented by 0.1 kg/day each week of test, up to a maximum of 2.3 kg/day. After completion of performance test, animals were transported from Edinburgh to Bristol. At Bristol, animals were fed the same diet and ration as at the end of the performance test, until slaughter 5 days later. Pigs were then moved to the slaughter-house and lairaged overnight to allow them to settle. Food was withdrawn no more than 12 h before slaughter and water was freely available. Muscle and eating quality traits were measured in the loin joints from the right hand side of each carcass (Cameron, Nute, Brown, Enser & Wood, 1999b). The longissimus dorsi muscle maximum depth (measured on the cut surface at the last rib) and the depth of subcutaneous fat, above the maximum muscle depth, were measured on the right-hand side of the cold carcass, but only in the DFI, LFC and LGA lines and the control line. Each animal was assessed for abnormal
N.D. Cameron et al. / Meat Science 54 (2000) 147±153
odour of cooked meat (1=low and 8=high intensity) by a sensory panel (Wood, Nute, Fursey, & Cuthbertson, 1995). Carcass subcutaneous fat weight was predicted from data on the dissected fore-loin joint, using a prediction equation determined from carcass data in a previous generation (Cameron & Curran, 1995): Carcass subcutaneous fat weight 3
Cwt Jwt i Jwti i1
where Cwt and Jwt are the carcass and fore-loin joint weights and Jwti are the weights of lean, subcutaneous fat and bone in the fore loin joint. The proportion of variation in carcass subcutaneous fat weight accounted for by the regression equation was 0.87. 2.2. Androstenone, skatole and indole measurement and sensory assessment of fat Fat samples were obtained from over the longissimus dorsi muscle. Androstenone concentration in fat was measured using the procedure described by De Brabander, Verbeke, Dirinck and Casteels (1985), with skatole and indole concentrations determined by the Likens-Nickerson method (Annor-Frempong, Nute, Whittingdon & Wood, 1997a). For sensory analysis, the fat samples were prepared as described by AnnorFrempong, Nute, Whittingdon and Wood (1997b). Brie¯y, 11±13 g samples of fat, taken from over the longissimus dorsi muscle, were individually heated for up to 8 min in Tecam heating blocks set at 100 C to a maximum of 65 C and then placed in 120 ml amber glass screw-top bottles. The sensory panel consisted of 10 females, aged 30±60 years, and each panellist was situated in an individual, well-ventilated booth. The panellists were selected for their ability to repeatedly distinguish between samples diering in fat content of androstenone and skatole. Four fat samples were given to each panelist in each session, with four sessions per day and 30 min between sessions. Each sample was assessed for intensity of abnormal odour, androstenone odour and skatole odour. The odour intensities were assessed on eight point scales (1=low and 8=high intensity). 2.3. Statistical analysis The genotype with nutrition interaction was estimated by residual maximum likelihood (REML) analyses using the REML algorithm of Genstat 5.3 Committee (1993). The model included ®xed eects of selection line, diet, feeding regime and two-way interactions between
149
selection line, diet and feeding regime, with batch included as a random eect. The three-way interaction of selection line, diet and feeding regime was initially included in the model to determine if the selection line with diet interaction was feeding regime dependent. For all traits, the three-way interaction was not statistically signi®cant and was deleted from the model. Cold carcass weight was included in the model as a covariate, as the experiment was designed to measure animals on a ®xed weight basis. Log transformations of fat content of androstenone, skatole and indole were required, as the traits were positively skewed. Throughout the text, all eects which are dierent at the 0.05 signi®cance level are referred to as being signi®cant. The probabilities of having a fat androstenone or skatole concentration above the thresholds of 1 or 0.25 mg/g, respectively, (Bonneau, 1997) were determined by analysing the binomial traits for each animal, denoting concentrations below or above the thresholds. The binomial traits were analysed using a generalised linear mixed model (Genstat 5.3 Committee, 1993). The model had a binomial error and a logit link (Schall, 1991; Welham, 1993) and included selection line, diet and feeding regime as ®xed eects, the two-way interactions between ®xed eects and batch as a random eect. Sensory panel scores for abnormal odour, androstenone odour and skatole odours were not transformed for analysis, as the precision of estimated betweenselection line dierences was not increased after transformation of sensory panel scores to standardised normal deviates (Cameron et al., 1999b). 3. Results 3.1. Growth and carcass traits Dierences between the selection and control lines, dietary and feeding regime eects on carcass traits are presented in Table 2. Between-selection line variation in cold carcass weight resulted from dierences in weight of transportation to Bristol, as when that weight was included in the model as a covariate for cold carcass weight, there were no dierences between the selection lines. The control line was intermediate to the LFC and DFI selection lines for both carcass content and growth rate of subcutaneous fat. The LGA and LGS lines had similar rates of subcutaneous fat growth as the control line. Diet A resulted in signi®cantly greater subcutaneous fat depth and higher predicted carcass content and growth rate of fat than diets C and E, which were similar. Predicted carcass content and growth rate of fat were also signi®cantly higher for ad libitum than restricted fed boars. There were no statistically signi®cant genotype with nutrition interactions for carcass traits.
150
N.D. Cameron et al. / Meat Science 54 (2000) 147±153
Table 2 Mean values for boar taint traits of the four selection lines (lean growth rate with ad-libitum, LGA or restricted, LGS feeding; lean food conversion, LFC; daily food intake, DFI), relative to the control line, of the three diets and two feeding regimes Genotype
Diet
Feeding regime
Control DFIa
LFC
LGA
LGS
s.e.d.
A
C
E
s.e.d.
ad libb restr
s.e.d.
Cold carcass weight (kg) Subcutaneous fat depth (mm) Predicted carcass sub. fat (g/kg) Pred. sub. fat growth rate (g/day)
67.9 12.1 255 212
7.7 ÿ0.8 46 58
0.5 ÿ4.4 ÿ53 ÿ49
ÿ5.6 ÿ5.2 ÿ34 9
0.3 ± ÿ30 ÿ15
3.6 1.7 18 17
69.2 12.5 282 246
68.0 8.2 220 204
68.3 7.7 220 189
1.9 1.0 10 15
69.6 9.4 254 244
67.4 9.6 228 183
1.3 0.8 7.4 10
Log androstenone Log skatole Log indole
0.15 ÿ2.29 ÿ3.73
ÿ0.18 0.11 0.17
ÿ0.05 ÿ0.55 ÿ0.01
0.05 0.09 ÿ0.03
ÿ0.32 ÿ0.53 0.06
0.23 0.45 0.26
0.12 ÿ2.42 ÿ3.56
0.03 ÿ2.45 ÿ3.66
0.01 ÿ2.52 ÿ3.85
0.15 0.32 0.20
0.21 ÿ2.20 ÿ3.52
ÿ0.11 ÿ2.73 ÿ3.86
0.11 0.23 0.15
0.51 0.0
0.05 0.2
0.10 0.4
ÿ0.23 ÿ1.0
ÿ0.24 ÿ1.0
1.4
0.44 ÿ0.3
0.54 0.2
0.34 ÿ0.7
0.8
0.62 0.5
0.27 ÿ1.0
0.6
0.09 ÿ2.3
0.08 0.8
0.02 0.2
0.17 1.3
ÿ0.08 ÿ2.3
1.4
0.10 ÿ2.2
0.13 ÿ1.9
0.11 ÿ2.1
0.07 ÿ2.6
0.7
4.42 2.89 5.35 3.04
ÿ0.25 0.18 ÿ0.03 0.06
ÿ0.43 0.36 ÿ0.21 0.17
ÿ0.28 0.56 ÿ0.25 0.42
ÿ0.04 ÿ0.38 ÿ0.02 ÿ0.11
0.35 0.33 0.37 0.30
4.14 3.00 5.15 3.19
4.21 3.02 5.12 3.14
4.21 3.02 5.25 3.12
4.24 3.04 5.24 3.18
P(androstenone>1 mg/g) Logit score=log[1/(1ÿprob)] P(skatole>0.25 mg/g) Logit score Fat: androstenone odour (1:weak) Fat: skatole odour (1:weak) Fat: abnormal odour (1:weak) Meat: abnormal odour (1:weak) a b
0.05 ÿ3.0 10 4.32 3.07 5.47 3.11
0.17 0.23 0.20 0.25
0.12 0.17 0.14 0.18
Dierence between the DFI, LFC, LGA and LGS genotypes and the control line. ad libitum and restricted feeding regimes.
3.2. Androstenone, skatole and indole Average fat contents of androstenone, skatole and indole were 1.05, 0.085 and 0.025 mg/g, respectively, but as log transformations were required for the statistical analyses, mean values of log transformed fat content of androstenone, skatole, indole are given in Table 2. There were no signi®cant dierences between the selection and control lines or diets for log transformed fat content of androstenone, skatole, indole. However, the probability of a skatole concentration of fat greater than 0.25 mg/g was signi®cantly higher in the DFI and LGA selection lines than in the LGS line. Ad libitum fed boars had signi®cantly higher log transformed fat content of androstenone, skatole and indole than restricted fed boars and the probability of an androstenone concentration greater than 1 was signi®cantly higher. There were signi®cant diet with feeding regime interactions for log transformed fat contents of skatole and indole, essentially due to signi®cantly higher log transformed fat contents with ad libitum feeding of diet E compared to restricted feeding (skatole: ÿ1.94 vs ÿ3.06, s.e.d. 0.43; indole: ÿ3.44 vs ÿ4.22, s.e.d. 0.28). Ad libitum feeding of diets A and C resulted in higher log transformed fat contents of androstenone, skatole and indole, but the within-diet contrasts were not signi®cantly dierent from zero. 3.3. Sensory assessment of fat There were no signi®cant dierences between selection and control lines for sensory panel score for
androstenone odour, but the LFC and LGA selection lines had a signi®cantly higher sensory panel score for skatole odour than the LGS selection line (Table 2). Both diet and feeding regime had no signi®cant eect on sensory panel assessment. There was a signi®cant selection line with feeding regime interaction for skatole odour, with a higher sensory panel score for ad libitum fed than for restricted fed LGA boars (3.87 vs 2.77, s.e.d. 0.32). A statistically signi®cant selection line with diet interaction was detected for sensory panel score of androstenone odour, due to higher scores for the LGA selection line on diet C compared to diets A and E (4.40 vs 3.71 and 3.81, s.e.d. 0.32), with the converse for the LGS line (4.00 vs 4.47 and 4.85). 3.4. Correlations between traits Correlations between traits are presented in Table 3. Log transformed fat content of androstenone, skatole and indole were signi®cantly positively correlated, particularly androstenone and skatole (Fig. 1). Log transformed fat contents of androstenone and skatole were signi®cantly correlated with sensory panel score for skatole odour, but not with panel score for androstenone odour. Sensory panel scores for androstenone odour and skatole odour were both signi®cantly correlated with sensory panel score for abnormal odour, although the former two traits were essentially independent. Animals were classed as low androstenone (LA)-low skatole (LS), high androstenone (HA)-LS, and HA-high skatole using the thresholds of 1 and 0.25 mg/g for fat content of androstenone and skatole,
N.D. Cameron et al. / Meat Science 54 (2000) 147±153
151
Table 3 Correlations (100) between fat content of androstenone, skatole, indole and sensory panel assessmenta
Log androstenone (AND) Log skatole (SKA) Log indole (IND) Fat: Androstenone odour (O-AND) Fat: Skatole odour (O-SKA) Fat: abnormal odour (O-ABN) Meat: abnormal odour (M-ABN) Predicted carcass sub. fat (FAT) Predicted sub. fat growth rate a
AND
SKA
IND
O-AND
O-SKA
O-ABN
M-ABN
FAT
80a 36 6 37 29 40 41 48
41 9 46 46 60 53 52
18 10 34 23 28 33
14 62 10 6 7
56 32 0 2
40 30 28
40 46
90
s.e. of correlations =0.12.
for sensory panel scores for skatole odour (1.36, s.e. 0.46 and 0.20, s.e. 0.10) and abnormal odour (1.13, s.e. 0.38 and 0.17, s.e. 0.08). The linear and quadratic regression coecients of log transformed fat content of indole were signi®cantly negative (ÿ3.8, s.e. 1.8 and ÿ0.50, s.e. 0.24) as explanatory variables for sensory panel scores for skatole odour. The coecients of determination (R2) of the models for sensory panel scores for skatole and abnormal odours were only 0.30 and 0.33, respectively, with no change in the latter when log transformed fat contents of androstenone or indole or both were dropped from the model. None of the regression coecients were signi®cantly dierent from zero with sensory panel score for androstenone odour as the dependent variable, in line with the corresponding correlation coecients. Fig. 1. Log transformed fat contents of androstenone and skatole.
respectively. However, the relationship between fat content and sensory panel odour score for androstenone and skatole with animals classed as LA-LS, LA-HS, HA-LS, and HA-HS (Annor-Frempong et al., 1997b) could not be examined, as there were insucient numbers of observations in the LA-HS and HA-LS classes (see Fig. 1). Predicted carcass content and growth rate of fat were signi®cantly correlated with log transformed fat content of androstenone and skatole, but not with sensory panel score for androstenone odour or skatole odour. Relationships between sensory panel androstenone, skatole and abnormal odour scores with linear and quadratic log transformed androstenone, skatole and indole fat content were examined in multiple regression analyses. Regression coecients for log transformed fat content of androstenone were not signi®cantly dierent from zero for sensory panel androstenone, skatole or abnormal odour scores. Linear and quadratic terms of log transformed skatole fat content were statistically signi®cantly dierent from zero as explanatory variables
4. Discussion The study was performed prior to the availability of results from associated meat quality studies using animals from the lean growth experiment. In retrospect, the general lack of signi®cant dierences in fat content of androstenone or in sensory panel assessment of boar taint traits in fat between the selection lines or diets for boars slaughtered at 90 kg was not unexpected. Firstly, there were no signi®cant dierences in eating quality between the selection lines, when performance tested on a single diet (Cameron et al., 1999b) or on the same three diets as in the current study (Cameron et al., 1999c). Secondly, it was unlikely that boars in the current study had reached sexual maturity as the average weight at ®rst oestrus of gilts from the selection lines was 124 kg, with ®rst oestrus determined from weekly plasma progesterone assay, (Cameron, Kerr, Garth & Sloan, 1999a). However, the study did detect a signi®cantly lower sensory panel score for skatole odour of fat in the LGS line compared to the LFC and LGA lines, such that the probability of a greater fat content of skatole than the threshold of 0.25 mg/g was lower in
152
N.D. Cameron et al. / Meat Science 54 (2000) 147±153
the LGS line. Therefore, selection for lean growth rate with performance testing on a restricted feeding regime, to produce animals of high genetic merit for ecient lean growth rate, may be associated with a lower incidence of boar taint, with regard to fat content of skatole. LundstroÈm et al. (1994) reported that selection for high lean growth on a high protein diet (185 g/kg) resulted in a lower fat content of skatole than selection on a low protein diet (131 g/kg). In the current study, there was no eect of diet on fat content of skatole given the range of dietary protein contents of 157, 210 and 262 g/kg, which was substantially greater than in the studies of LundstroÈm et al. (1988) (144 and 163 g/ kg), LundstroÈm et al. (1994), Andersson et al. (1997) (138 and 141 g/kg) and Nold, Romans, Costello, Henson and Libal (1997) (120±170 g/kg), which also reported no consistent signi®cant eect of diet. Therefore, dierences in fat content of skatole in the selection lines of LundstroÈm et al. (1994) may not have been due to dierences in dietary protein content - a point also made by LundstroÈm et al. (1994). Further, the selection lines of LundstroÈm et al. (1994) diered to the same extent in fat content of skatole in generations one, two and four, suggesting that the higher frequency of the allele for high fat content of skatole in the line selected on a low protein diet was due to serendipity, given the assumption of LundstroÈm et al. (1994) that a single gene exists for fat content of skatole. The diet with feeding regime interaction for fat content of skatole and indole, due to signi®cantly higher fat contents with ad libitum feeding than with restricted feeding for diet E than for diets A and C, may primarily be a consequence of higher intake of dietary tryptophan, as skatole and indole are formed from intestinal degradation of tryptophan (Claus et al., 1994). Diet E was included in the study to generate a sucient range of diets, such that dietary eects could be detected. If diet E had not been included in the study, then the diet with feeding regime interaction on log transformed fat contents of androstenone, skatole or indole would not have been detected. In previous studies, the range of diets has been too small that it was unlikely that a dietary eect could have been detected, if such an eect existed. Therefore, it is important that nutritional studies include diets which dier suciently to generate the necessary experimental power to detect dietary dierences. Despite substantial dierences in carcass fat content and fat growth rate between the DFI, control and LFC selection lines, there were no signi®cant dierences in log transformed fat content of androstenone, skatole and indole. Similarly, there were signi®cant dierences in carcass fat content and fat growth rate between diets, but they were not accompanied by dierences in log transformed fat content of androstenone, skatole and indole. Therefore, neither the rate of fat deposition nor
the carcass content of fat per se contribute consistently to the fat content of androstenone, skatole and indole, despite statistically signi®cant positive correlations between carcass content and growth rate of fat with log transformed fat content of androstenone, skatole and indole. Log transformed fat content of androstenone and indole provided no additional information on sensory panel score for abnormal odour in the current study, that was not provided by log transformed fat content of skatole. In contrast, Annor-Frempong et al. (1997b) reported that linear and quadratic terms for fat content of androstenone, skatole and indole proportionally accounted for 0.76 of the variation in sensory panel score for abnormal odour. The diering conclusions in the two studies may have stemmed from dierences in the observed variation in fat content of androstenone and in the statistical analyses. The minimum fat content of androstenone in the current study was 0.37 mg/g, while gilts were included in the Annor-Frempong et al. (1997b) study, such that half the animals had a fat content of androstenone of less than 0.30 mg/g. Despite the skewed distributions for fat content of androstenone, skatole and indole, the data were not transformed in the Annor-Frempong et al. (1997b) study, such that standard errors of the regression coecients may have been under-estimated resulting in false positive conclusions. Studies designed to determine the predictive value of fat content of androstenone, skatole and indole for sensory panel assessment of abnormal odour should dierentiate between boars and gilts and should account for the non-normal distributions of fat content of androstenone, skatole and indole. Based on the numbers of animals in the study, the power of the experiment appears to be limited for detection of statistically signi®cant dierences in boar taint traits between the selection lines and the control line. The experimental power was determined using the formula of Hill (1978) and assumptions regarding the estimation of drift and sampling variances are given by Cameron, et al. (1999c). To achieve an experimental power of 0.80 for detection of signi®cant correlated responses in boar taint traits relative to the control line, the parameter rA h needed to exceed 0.18, where rA is the genetic correlation of a boar taint trait with the selection criterion and h2 is the heritability of the boar taint trait. For example, a genetic correlation of 0.4 and a heritability of at least 0.2 would be sucient, which was thought reasonable when the experiment was designed. Even if the experiment was quadrupled in size, the required minimum value of rA h would only reduce to 0.16, such that a genetic correlation of 0.3 would suce. The experimental power was enhanced by measuring the response in selection lines relative to the control line, when the lines were derived from the same base population, rather than from having a high number of
N.D. Cameron et al. / Meat Science 54 (2000) 147±153
animals per selection line. However, it would be instructive to compare standard errors of dierences in the current study with those in previous studies, but literature information on the variation of studied traits is limited, as several recent studies (Bonneau, le Denmat, Vaudelet, Veloso Nunes, Mortensen & Mortensen, 1992; Xue et al., 1996; Annor-Frempong et al., 1997b) have not reported either standard deviations or standard errors of dierences between ``treatments''. In conclusion, there was no evidence of selection strategy speci®c responses in fat content of androstenone, skatole or indole, traits associated with boar taint. However, selection for rate of lean growth on a restricted feeding regime resulted in a lower sensory panel score for skatole odour than selection for either rate or eciency on lean growth on ad libitum feeding. The higher fat content of skatole and indole with ad libitum feeding compared to restricted feeding of a diet with a high ileal digestible lysine content may have primarily been a consequence of higher intake of dietary tryptophan. While fat content of androstenone and skatole could be used as predictors of sensory panel odour score for skatole, they would not be useful predictors of androstenone odour. Acknowledgements The project was funded by the Ministry of Agriculture, Fisheries and Food. The diets were designed by M. MacLeod, Roslin Institute. References Andersson, K., Schaub, A., Andersson, K., LundstroÈm, K., Thomke, S., & Hansson, I. (1997). The eects of feeding system, lysine level and gilt contact on performance, skatole levels and economy of entire male pigs. Livestock Production Science, 51, 131±140. Annor-Frempong, I. E., Nute, G. R., Whittington, F. W., & Wood, J. D. (1997a). The problem of taint in pork II. The in¯uence of skatole, androstenone and indole, presented individually and in combination in a model lipid base, on odour perception. Meat Science, 47, 49±61. Annor-Frempong, I. E., Nute, G. R., Whittington, F. W., & Wood, J. D. (1997b). The problem of boar taint in pork III. Odour pro®le of pork fat and the interrelationships between androstenone, skatole and indole concentrations. Meat Science, 47, 63±76. Bonneau, M., le Denmat, M., Vaudelet, J. C., Veloso Nunes, J. R., Mortensen, A. B., & Mortensen, H. P. (1992). Contributions of fat androstenone and skatole to boar taint: I. Sensory attributes of fat and pork meat. Livestock Production Science, 32, 63±80. Bonneau, M. (1997). Boar taint in entire male pigs Ð a review. Pig News and Information, 18, 15N±18N. Cameron, N. D. (1994). Selection for components of ecient lean growth rate in pigs. 1. Selection pressure applied and direct responses in a Large White herd. Animal Production, 59, 251±262.
153
Cameron, N. D., & Curran, M. K. (1995). Responses in carcass composition to divergent selection for components of ecient lean growth rate in pigs. Animal Science, 61, 347±359. Cameron, N. D., & MacLeod, M. G. (1997). Genotype with nutrition interaction for performance test traits in pigs selected for lean growth rate. Proceedings of the British Society of Animal Science Annual Meeting. Paper 29. Cameron, N. D., Kerr, J. C., Garth, G. B., & Sloan, R. L. (1999a). Genetic and nutritional eects on age at ®rst oestrus of gilts selected for components of ecient lean growth rate. Animal Science, 69, 93±103. Cameron, N. D., Nute, G. R., Brown, S. N., Enser, M., & Wood, J. D. (1999b). Meat quality of Large White pig genotypes selected for components of ecient lean growth rate. Animal Science, 68, 115±127. Cameron, N. D., Penman, J. C., Fisken, A. C., Nute, G. R., Perry, A. M., & Wood, J. D. (1999c). Genotype with nutrition interactions for carcass composition and meat quality in pig genotypes selected for components of ecient lean growth rate. Animal Science, 69, 69±80. Claus, R., Weiler, U., & Herzog, A. (1994). Physiological aspects of androstenone and skatole formation in the boar Ð a review with experimental data. Meat Science, 38, 289±305. De Brabander, H. E., Verbeke, R., Dirinck, P., & Casteels, M. (1985). Boar taint in Belgian pigs in relation to the androstenone content. Proc. 31st Europ. Meet. Res. Workers. Varna, 778±781. Fouilloux, M. N., Le Roy, P., Gruand, J., Renard, C., Sellier, P., & Bonneau, M. (1997). Support for single major genes in¯uencing fat androstenone level and development of bulbo-urethral glands in young boars. Genetics Selection Evolution, 29, 357±366. Genstat 5.3 Committee (1993). Genstat 5.3 reference manual. Oxford: Clarendon Press. Hill, W. G. (1978). Design of selection experiments for comparing alternative testing regimes. Heredity, 41, 371±376. LundstroÈm, K., Malmfors, B., Malmfors, G., Stern, S., Petersson, H., Mortensen, A. B., & Sùrensen, S. E. (1988). Skatole, androstenone and taint in boars fed two dierent diets. Livestock Production Science, 18, 55±67. LundstroÈm, K., Malmfors, B., Stern, S., Rydhmer, L., Eliasson-Selling, L., Mortensen, A. B., & Mortensen, H. P. (1994). Skatole levels in pigs selected for high lean tissue growth rate on dierent dietary protein levels. Livestock Production Science, 38, 125±132. Nold, R. A., Romans, J. R., Costello, W. J., Henson, J. A., & Libal, G. W. (1997). Sensory characteristics and carcass traits of boars, barrows and gilts fed high- or adequate-protein diets and slaughtered at 100 or 110 kilograms. Journal of Animal Science, 75, 2641± 2651. Schall, R. (1991). Estimation of generalised linear models with random eects. Biometrika, 78, 719±727. Valaja, J., Suomi, K., Alaviuhkola, T., & Immonen, I. (1992). Eect of dietary ®sh-meal on the palatability and fatty-acid composition of pork. Agricultural Science in Finland, 1, 21±26. Van Lunen, T. A., & Cole, D. J. A. (1996). The eect of lysine/digestible energy ratio on growth performance and nitrogen deposition of hybrid boars, gilts and castrated male pigs. Animal Science, 63, 465± 475. Welham, S. J. (1993). The GLMM procedure. Genstat 5, Procedure library manual release, 3(1), 187±192. Willeke, H. (1993). Possibilities of breeding for low 5a-androstenone content in pigs. Pig News and Information, 14, 31N±33N. Wood, J. D., Nute, G. R., Fursey, G. A. J., & Cuthbertson, A. (1995). The eect of cooking conditions on the eating quality of pork. Meat Science, 40, 127±135. Xue, J. L., Dial, G. D., Holton, E. E., Vickers, Z., Squires, E. J., Loy, Y. P., Godbout, D., & Morel, N. (1996). Breed dierences in boar taint: relationships between tissue levels of boar taint compounds and sensory analysis of taint. Journal of Animal Science, 74, 2170±2177.