Station-test

In Switzerland, approximately 3000 growing pigs are performance tested per year. Purebred animals of the Large White dam line (LW_DL), Swiss Landrace (SLR) and Large White Sire line (LW_SL) make up more than 80% of all test animals. Usually, a test group consists of two full-sibs (one female and one male castrate). Animals are housed in 12 barns, each with eight pens and about 10 animals of both sexes per pen. Electronic feed dispensers are used to feed the animals ad libitum. Two different diets are used during the test. At first, each pig obtains approximately 80 kg of a starter feed (19% CP, 13.3 MJ digestible energy (DE), 10 g/kg lysine, 8 g/kg Ca, 5 g/kg P). Afterwards, a finisher feed (15% CP, 12.7 MJ DE, 9 g/kg lysine, 8 g/kg Ca, 5 g/kg P) is used. The test starts at 30 kg live weight and ends with approximately 103 kg. Every week, 40 to 70 animals are slaughtered in a small abattoir at the test station and the left half of the carcass is dissected.

Average daily gain (ADG) from 30 to 103 kg describes the growth of animals. Percentage of premium cuts (PPC) specifies the leanness of carcasses. PPC is defined as the weight of ham, loin and shoulder without fat layer proportional to carcass weight. A full description of the Swiss station-test is given in the annual report of SUISAG (2006).

Examination of bones

Osteochondral lesions were already observed at the Swiss test station in the 1980s and early 1990s (Schwörer et al., Reference Schwörer, Lorenz and Rebsamen1991). In 2002, more weight was placed on daily gain in the overall breeding goal. SUISAG decided to monitor the occurrence of osteochondral lesions, because unfavourable genetic correlations with daily gain may exist.

Since 2002, osteochondral lesions have been examined at the following bones and joints: head of humerus (HK), condylus medialis humeri (CMH), condylus lateralis humeri (CLH), radius and ulna proximal (RUP), head of femur (FK), distal epiphyseal cartilage of ulna (DEU) and condylus medialis femoris (CMF). Figure 1 shows the corresponding positions at the carcass. At first, a trained person examines the surface and shape of these bones at the joints. Afterwards each bone is sawed with a band saw and the state and shape of the cartilage and the DEU is examined at the cutting surface. Scores from 1 to 4 are used to record the extent of the osteochondral lesion. A score of 1 denotes ‘no visible osteochondral lesion’. Scores 2 to 4 denote ‘mildly to severely affected’, respectively. Osteochondral lesions at the DEU and CMF are scored from 1 to 6, because these lesions seemed to be more variable. More details concerning the Swiss examination of osteochondral lesions are given by Schwörer et al. (Reference Schwörer, Häni and Blum1985).

Figure 1 Examinations of osteochondral lesions at the Swiss test station (Schwörer et al., Reference Schwörer, Häni and Blum1985).

Linear description of exterior

A linear description system (LD) has been used to evaluate exterior characteristics of Swiss gilts and boars (on-farm test) and all station-tested pigs since 2000. Five traits are used to describe the rear and fore extremities: X-O posture at rear legs (XORL), side-view angle of rear legs (SVARL), angle of pasterns at rear legs (APRL), size of the inner claw in comparison to the outer claw at rear legs (SICRL) and side-view angle of fore legs (SVAFL). These five traits are scored from 1 to 7 with score 4 as the optimal phenotype and scores 3 to 1 and 5 to 7 as more and more extreme and therefore negative phenotypes. Scores from 4 to 7 are used to describe the regularity at the loin (LOIN). Score 4 denotes ‘regular’ and scores 5 to 7 denote that the animals are increasingly narrow at the loin in comparison to the whole body. Finally, scores from 4 to 7 are used to evaluate the locomotion (LOC) of the animal. Score 4 denotes ‘regular locomotion’ and scores 5 to 7 denote a more and more lurching or stiff locomotion. Figure 2 shows a scheme of the linear description system. Tarrés et al. (Reference Tarrés, Bidanel, Hofer and Ducroq2006) recently showed that gilts with an optimum exterior – i.e. only scores of 4 at the linear description – stayed longer in the farms compared with gilts with a suboptimal exterior.

Figure 2 Scheme of the Swiss linear description system (without teat description) (some drawings from Van Steenbergen, Reference Van Steenbergen1990).

The present study includes exterior examinations of station-tested animals, which were performed approximately 4 weeks to 1 week before slaughtering. Seven trained persons described the animals during the evaluation period. The length of the carcass (CL) of every animal was measured from the pubis bone to the cervical vertebra after slaughtering. A detailed report of the applied linear description system and the genetic evaluation of exterior traits in Switzerland is given by Hofer (Reference Hofer2004).

Data

This study includes 9511 performance-tested animals of the purebred lines LW_DL, SLR and LW_SL that were slaughtered between January 2002 and December 2005 (Table 1). Table 2 presents the average fattening and carcass performance of the station-tested animals. The test animals are progeny of 1051 sires and 19 988 animals are included in the pedigree file.

Table 1 Distribution of station-tested pigs

^† LW_DL = Large White dam line; SLR = Swiss Landrace; LW_SL = Large White sire line.

Table 2 Mean and distribution of performance traits (n = 9511)

^† LW = live weight at slaughtering; ADG = average daily gain (30 to 103 kg); FCR = feed conversion ratio; FI = average daily feed intake; CL = length of carcass; PPC = percentage of premium cuts.

About 27% of all slaughtered animals were examined for osteochondral lesions (n = 2622; Table 3). The number of observations was slightly reduced at RUP and DEU, because these bones were used for another project. Data from linear description were available for nearly all animals. Approximately 50 test animals had no or incomplete exterior data due to different reasons (e.g. clinical lameness at description day, etc.).

Table 3 Distribution of osteochondral lesions within breedsFootnote ^†

^† For abbreviations see Figure 1.

Statistical analysis

Genetic parameters were estimated with univariate and multivariate linear mixed animal models using VCE5 (Kovač et al., 2002). Due to the categorical characteristic and the very skewed distribution, HK, CLH, RUP and FK were excluded from the final statistical analyse. Preliminary univariate estimations showed nearly no genetic variance for these four traits. CMH, DEU and CMF exhibited a prevalence of more than 10% and remained in the statistical analyses with linear models. The heritabilities on the underlying liability scale might be underestimated according to the formula of Robertson and Lerner (Reference Robertson and Lerner1949) due to the use of linear models, especially for CMH. But according to Gianola (Reference Gianola1984) the genetic correlations with other continuously distributed traits are almost not affected. Estimations of heritabilities and genetic correlations of the 13 traits were performed in 14 separate runs. It was not possible to run more than seven traits simultaneously. The results of all runs were averaged. Slightly different linear mixed models were used for different traits.

ADG:

PPC, CL:

XORL, SVARL, APRL, SICRL, SVAFL, LOIN, LOC:

CMH, DEU, CMF:

with:

y

observation of test animal p (p = 1, 2, …, 9511)

μ

overall mean

BREED_h

fixed effect of breed (h = LW_DL, SLR, LW_SL)

BARN_PER_i

fixed effect of barn × period combination (i = 1, 2, …, 149)

SL_DAY_j

the fixed effect of slaughter day (j = 1, 2,…, 213)

SL_PER_k

fixed effect of slaughter period (k = 1, 2, …, 8)

LD_PE_l

fixed effect of the person performing theLD (l = 1, 2, …, 7)

SEX_m

fixed effect of sex (m = female, castrate)

b_mLW

partial linear regression on live weight within sex

farm_y_n

random effect of farm of origin × year (n = 1, 2, …, 249)

litter_o

random effect of the litter (o = 1, 2, …, 4763)

animal_p

random effect of the animal (p = 1, 2, …, 9511)

e

random error of test animal p (p = 1, 2, …, 9511).

All test animals that were fattened in the same barn and period were used as a contemporary group for ADG and exterior traits. Eight slaughter periods (6 months) were used as a contemporary group for osteochondral lesions traits, because only approximately 10 to 15 pigs were examined for OC lesions at each week after slaughtering. The test animals were born at 78 farms within 4 years and belong to 4763 different litters. ADG values were pre-corrected to a constant live weight at slaughtering (103 kg). Therefore, no regression on live weight was included in the model concerning this trait. No correction for live weight was possible for exterior traits, because pigs were not weighed at the description day.

The statistical software S-PLUS (S-PLUS, 1999) was used to test the significance of fixed effects for osteochondral lesions (Table 4).

Table 4 Significance of fixed effectsFootnote ^†

^† For abbreviations see Figure 1 and the text. ***P≤0.001.

Heritabilities and correlations

At first, univariate analyses were performed to estimate the variance components of osteochondral lesion at CMH, DEU and CMF (Table 6). The additive genetic effect explains a noticeable proportion of the total phenotypic variance of osteochondral lesions. Hence, heritabilites for these traits ranged from h ² = 0.16 to 0.18. The corresponding standard errors were low.

Table 6 Variance components and heritabilities (h ²) of osteochondral lesions at different positions of the carcass (univariate estimation)Footnote ^†

^† c ² = ratio of litter variance, s.e. = standard error of h ² and for other abbreviations see Figure 1.

The random effect farm of origin × year combination explains nearly no variance. This indicates that the farm of origin does not influence the occurrence of osteochondral lesions at the end of the station test significantly. Concerning osteochondral lesions at CMH, the litter effect explains a minor proportion of the total variance. This means that pigs of the same litter show more frequently OC lesions than one would expect due to the identical parentage. Environmental effects, e.g. low milk production of the nursing dam, could be a potential explanation for this observation. But on the other hand no noticeable litter effect was observed for DEU and CMF.

Kadarmideen et al. (Reference Kadarmideen, Schwörer, Ilahi, Malek and Hofer2004) estimated very low heritabilities for osteochondral lesions using linear mixed sire models except for RUP and CMF. But the authors used threshold models too. Heritabilities obtained from these threshold models ranged between h ²=0.00 (FK) and h ² = 0.42 (CMH). Transformed to the ‘observed scale’, these heritabilities are in good agreement with the values of the present study except for DEU, which was not significantly different from zero in the study of Kadarmideen et al. (Reference Kadarmideen, Schwörer, Ilahi, Malek and Hofer2004).

Jørgensen and Andersen (Reference Jørgensen and Andersen2000) estimated heritabilities of OC at six different positions of the carcass. The values ranged from h ² = 0.08 to 0.39 and varied sometimes between Danish Yorkshire and Landrace breeds at the same position of the carcass. On average, the heritabilities in Danish populations were higher than in the present study. Differences between Danish and Swiss populations might be a potential reason. Obviously, there were distinctions between Danish breeds. We did not estimate genetic parameters of the three Swiss breeds separately, because less than 500 SLR and SLW_SL test animals had OC data, respectively. Moreover, phenotypic variation of OC was more pronounced in Danish populations compared with the present study. Jørgensen and Andersen (Reference Jørgensen and Andersen2000) examined both sides of the animal radiologically for osteochondral lesions and analysed the average score statistically. The different methods of examination might have affected the incidence and variation of OC and could be another cause of different heritabilities.

Table 7 presents the results of the multivariate estimation of genetic parameters. Phenotypic correlations of osteochondral lesions to exterior and production traits were close to zero. Heritabilities of osteochondral lesions changed slightly due to the multivariate estimation in comparison to the univariate analyses. Linearly described exterior traits showed low heritabilities (0.10 to 0.20). But all values differed significantly from zero. CL and the PPC exhibited the highest heritabilities of all traits (0.61 and 0.62, respectively). Serenius et al. (Reference Serenius, Sevón-Aimonen and Mäntysaari2001) estimated lower heritabilities for leg weakness symptoms in comparison to our findings. The present results are in good agreement with Jørgensen and Andersen (Reference Jørgensen and Andersen2000).

Table 7 Heritabilities (diagonal, bold), phenotypic correlations (above diagonal) and genetic correlations (below diagonal) of multivariate analysesFootnote ^†

^† For abbreviations see Figures 1 and 2 and Table 2. Underlined values indicate notable genetic correlation is favourable (r_g ⩾ ± 0.15). Italic values indicate notable genetic correlation is unfavourable (r_g ⩾ ± 0.15).

Genetic correlations between osteochondral lesions at CMH, DEU and CMF were low (r_g = −0.14 to 0.13). This indicates that the occurrence of osteochondral lesions at the three positions does not seem to be controlled by identical genes and should be considered as different traits. This conclusion agrees with Jørgensen and Andersen (Reference Jørgensen and Andersen2000) and Kadarmideen et al. (Reference Kadarmideen, Schwörer, Ilahi, Malek and Hofer2004).

The genetic correlations of osteochondral lesions to ADG and PPC were very low. Only the unfavourable genetic correlation between osteochondral lesions at CMF and ADG exceeded r_g = ±0.15 in the present study. Thus, genetic selection that favours leaner and faster growing pigs does not seem to increase the genetic predisposition to exhibit osteochondral lesions considerably in Swiss breeds. Our results are in agreement with Lundeheim and Rydhmer (1990), Jørgensen and Andersen (Reference Jørgensen and Andersen2000) and Kadarmideen et al. (Reference Kadarmideen, Schwörer, Ilahi, Malek and Hofer2004). Those authors estimated few unfavourable genetic correlations with moderate magnitude between OC and production traits. In the same way, Jørgensen (Reference Jørgensen1995) did not find different prevalences of OC between pigs fed restricted v. ad libitum although the pigs fed ad libitum grew faster. In contrast, Busch et al. (2006) reported that a high ADG was associated with an increased risk to exhibit OC.

Genetic correlations of osteochondral lesions at CMH, DEU and CMF to exterior traits showed low to medium values. Nearly all correlations that exceed r_g = ±0.15 were favourable concerning the actual Swiss breeding goal. Especially the occurrence of osteochondral lesions at DEU and the regularity of locomotion seem to be genetically linked (r_g = 0.54). Only the association between regularity at the loin and CMH was genetically unfavourable and exceeded r_g = ±0.15. Therefore, genetically favouring animals with regular locomotion and optimal exterior would reduce the occurrence of osteochondral lesions in the long-term especially at DEU.

In general, our results agree with results by Jørgensen (Reference Jørgensen2000) and Jørgensen et al. (Reference Jørgensen, Arnbjerg and Aaslyng1995). Those authors also found slight phenotypic associations between several leg weakness symptoms and pathological changes of bones and joints. Jørgensen and Andersen (Reference Jørgensen and Andersen2000) estimated some moderate genetic correlations between leg weakness symptoms and OC at the femoral and humeral condyles too.

Conclusion and implications

At HK, CLH, RUP and CMF even mildly osteochondral lesions are rare in Swiss breeds. More than 10% of the pigs exhibited mostly mildly osteochondral lesion at the CMH, DEU and CMF, respectively. Thus, based on the station-tested pigs osteochondral lesions do not seem to be a severe problem in Swiss breeds at the moment. The occurrence of osteochondral lesions at CMH, DEU and CMF is heritable (h ² = 0.16 to 0.18). Therefore, it is possible to reduce the prevalence of osteochondral lesions by selection in principle.

Direct selection would require examination of all tested pigs for OC lesions, which is too costly. An indirect selection based on exterior traits is difficult due to low heritabilities of OC and exterior traits and low relative weight of exterior in the overall breeding goal. But the low genetic correlations between the most important production traits (ADG and PPC) and osteochondral lesions as well as the favourable genetic correlations with Swiss exterior traits suggest that the prevalence of OC will not increase, if gilts and boars are selected according to the Swiss breeding goal.

Nevertheless, SUISAG will continue the examinations of a random sample of station-tested pigs after slaughtering to observe the further development of osteochondral lesions in the Swiss breeds directly.