In this study, we evaluated conformation traits describing leg and feet condition, osteochondrosis score and bone mineral density, which are important in selection to reduce the risk of leg weakness in pigs. However, the genetics of LW-related traits is complex [12, 31]. A number of factors are known to influence the development of LW, such as nutrition imbalance, high body weight, rapid growth rate, bone and joint diseases, bad body and leg structure, and mechanical stress [11, 13]. Moreover, it has been reported that the degree of LW and OC may be related to the breed and sex of animals . However, in our study there was no effect of gender on LW-related traits, which implies that frequencies of LW and OC vary and depend on the genetic background of the animals . It has been reported that the Duroc pure breed shows the highest incidence of OC compared to other European pig breeds (Pietrain, Landrace and Yorkshire) . Our data suggest that the unfavourable QTL allele for OC originates from both Duroc and Pietrain breeds (i.e. two QTL originated from the Duroc and three from the Pietrain) (Table 3) and that in Duroc and Pietrain crossbred animals, the fore legs are less susceptible to OC than the rear legs. Andersson-Eklund et al.  have also reported lower OC incidences in the humerus than in the femur in a Wild boar × Large White population. In addition, our data show that the frequency of OC is high (31.05%) at CMF, which agrees with a previous report of 30.0% by Kadarmideen et al. .
QTL analyses for leg weakness and bone-related traits have been performed in different pig breeds, including Landrace purebred , White Duroc × Erhualian [19, 21], Large White × Meishan , Duroc × Landrace and Duroc × Large White crossbred , and Wild boar × Large White . To the best of our knowledge, our study is the first to map QTL for LW-related traits in a Duroc and Pietrain intercross. We have identified 11 QTL some of which being novel and some confirming previous studies [17–21, 34], as described in the next section. However, large confidence regions were obtained in this experiment, which represents a common problem in QTL studies and hampers the comparison of QTL results and their interpretation in terms of causative genes, since large confidence intervals can contain many potential candidate genes .
In this study, a QTL for FFS was detected on SSC1 at 166 cM. QTL for the same trait have been reported at 89 cM in a Landrace purebred  and at 52 cM in a Large White × Meishan intercross  on the same chromosome. The dominant QTL for FFS found on SSC16 at 36 cM is close to a previously reported dominant QTL at 27 cM for rear leg score . The QTL identified for RLS on SSC6 and SSC18 are new and do not overlap with any previous study. A QTL associated with rear leg score was observed on SSC6, close to marker SW193 (SSC6q2.1), where the gene for transforming growth factor-beta 1 (TGFβ1) is located . This gene is an important candidate for LW-related traits since TGFβ1 is a potent regulator of cell proliferation and influences the size and shape of the limb . We identified a QTL for the OC score at HH on SSC2 at 14 cM, while Christensen et al.  have reported QTL associated with cartilage thickening of the medial part of condylus humeri at 15 cM on the same chromosome. In addition, a QTL with dominance effect identified for the OC score at HH on SSC6 at 61 cM is located close to previously reported QTL for depression of the proximal edge of the radius at 51 cM  and for physis score at 75 cM . QTL for HH on SSC3 at 13 cM and for CMH on SSC14 at 0 cM are new QTL (Figure 1). Interestingly, the QTL for CMF on SSC10 at 70 cM is close to a previously identified QTL regions at 75 cM for OC lesion in the subchondral bone of the medial part of condylus humeri and at 83 cM for fissure between cartilage and bone in pigs . The QTL on SSC2 at 0 cM, close to marker SW2443 (SSC2p17), was the only QTL detected for BMC. One of the highest linkage associations, reaching a 5% GW significance, was found on SSC3 at 71 cM for BMD. A potential candidate gene in this chromosomal region is the follicle-stimulating hormone receptor (FSHR) gene, which directly regulates bone mass . These QTL for BMC and BMD are novel and do not overlap with previously reported QTL.
Most of the identified QTL show large dominance effects rather than additive affects (Table 3). It is important to note that the transformation done on the leg score traits in this study did not change the identified QTL regions since the interval mapping results for these traits using the original score ranging from 1-5 or the scale 0-2 were the same. This implies that the transformation done on the leg score is not the reason for over-dominance in this experiment.
In another QTL study in the same population, 31 of 71 QTL for growth, fatness, leanness and meat quality traits have also shown high dominance effects , as well as QTL for immune traits . Lee et al.  have also reported that most QTL for LW-related traits in a Large white × Meishan cross show dominance. In addition, using principal components analysis, Andersson-Eklund et al.  have identified a QTL for OC with a significant and large effect of over-dominance. Therefore, the results from this study and from previous studies reported in the literature [17–20, 34] suggest that dominance plays a role in the genetic control of LW-related traits.
Most of the traits analysed in this study are categorical rather than normally distributed. Previous studies have shown that the QTL analysis method  used is suitable for categorical traits, with little loss of power [19, 20]. The low heritability of these traits indicates that they may be complex traits and may be under a polygenic control primarily by non-additive gene action or affected by a major gene with Mendelian transmission . In this study, most of the QTL were identified as single-trait regions. This could be explained by the low phenotypic correlations observed between the traits in the population.