Effect of the Texel muscling QTL (TM-QTL) on spine characteristics in purebred Texel lambs

Abstract

Previous work showed that the Texel muscling QTL (TM-QTL) results in pronounced hypertrophy in the loin muscle, with the largest phenotypic effects observed in lambs inheriting a single copy of the allele from the sire. As the loin runs parallel to the spinal vertebrae, and the development of muscle and bone are closely linked, the primary aim of this study was to investigate if there were any subsequent associations between TM-QTL inheritance and underlying spine characteristics (vertebrae number, VN; spine region length, SPL; average length of individual vertebrae, VL) of the thoracic, lumbar, and thoracolumbar spine regions. Spine characteristics were measured from X-ray computed tomography (CT) scans for 142 purebred Texel lambs which had been previously genotyped. Least-squares means were significantly different between genotype groups for lumbar and thoracic VN and lumbar SPL. Similarly for these traits, contrasts were shown to be significant for particular modes of gene action but overall were inconclusive. In general, the results showed little evidence that spine trait phenotypes were associated with differences in loin muscling associated with the different TM-QTL genotypes.

1. Introduction

Walling et al. (2004) reported evidence of a quantitative trait locus (QTL) segregating in the United Kingdom's Texel sheep population which significantly increased longissimus dorsi (loin) muscle depth (by 1.15–2.00 mm, as measured ultrasonically over the third lumbar vertebra). With similar results (QTL effect of +2.57 mm) observed from an analysis including existing and new Texel family data, the QTL, later termed the Texel muscling QTL (TM-QTL), was further verified by Matika et al. (2006). Located on the distal end of the ovine chromosome 18 (OAR18) (Walling et al., 2004; Matika et al., 2006), the TM-QTL sits in the same region as the Callipyge (CLPG) and Carwell loci (Cockett et al., 1994; Nicoll et al., 1998), which are also known to affect carcass muscling. The CLPG mutation leads to greater muscle mass most pronounced in the hind quarters (loin, pelvis, leg) (Cockett et al., 1994; Koohmaraie et al., 1995; Jackson et al., 1997a,b; Freking et al., 2002), while carriers of Carwell exhibit a larger loin muscle area and weight (Nicoll et al., 1998; Masri et al., 2010).

Such QTL are of economic interest as there is the potential to utilise their effects through selection programmes to gain greater carcass value (e.g. reducing fat deposition and increasing lean meat production). In the case of the TM-QTL, the proportion of the high value loin cut may be increased. For example, two-dimensional measurements (estimated from cross-sectional computed tomography (CT) scans, taken at the fifth lumbar vertebra) describing loin depth, width, and area were found to be ∼0.5–11% greater in TM-QTL carrier lambs than non-carrier lambs (Macfarlane et al., 2010).

Moreover, taking the QTL's mode of inheritance into consideration allows the opportunity to exploit the TM-QTL more fully and appropriately in a commercial situation. Similar to CLPG (Cockett et al., 1994; Freking et al., 1998a), expression of the Texel muscling phenotype has been suggested to follow the complex parent-of-origin-dependent pattern of inheritance termed polar overdominance (Macfarlane et al., 2010; Matika et al., 2011), where heterozygous progeny that inherit a single copy of the allele from the sire exhibit the superior phenotype (Cockett et al., 1996). Indeed, Macfarlane et al. (2010) observed that the largest phenotypic effects of the TM-QTL were apparent in the lambs that had inherited a single copy of the TM-QTL from their sire and the wild type allele from their dam, with loin depth, width, and area measures ∼2–11% greater and loin weight ∼3–5% greater in these lambs than in the other three genotype groups (homozygote non-carriers, heterozygote carriers inheriting TM-QTL from the dam, homozygote carriers).

Essentially, muscle hypertrophy from TM-QTL segregation appears to be restricted to the loin muscle (Macfarlane et al., 2010), which is located along the length of the thoracolumbar (thoracic plus lumbar) spine region. Given that development of muscle and bone are closely linked, it is of further interest to investigate, across genotype groups, if the increased loin muscling is associated with any change to characteristics of the underlying spine section i.e. is there an effect on spine characteristics in relation to the pattern of TM-QTL inheritance? Freking et al. (1998b), for instance, found that the spinal column was significantly shorter in CLPG genotype lambs (−2.5 cm; when all animals were compared at the same carcass weight) and the carcasses were more compact in skeletal structure in comparison to normal genotype lambs. Given the proximity in chromosomal position of CLPG and TM-QTL, a similar effect may be expected for the TM-QTL. This is a particularly relevant point to assess in terms of a possible ‘trade-off’ between increasing loin muscle size (e.g. area, depth) but, in consequence, shortening the spinal column.

The thoracolumbar spine region, on which the loin is located, encompasses the ‘body’ (or trunk) vertebrae and the total length of this region (as with any spine region) is a product of the number and length of vertebrae which comprise it. Hence, the difference in body (and carcass) lengths observed from individual to individual is contributed to the variation in these vertebral factors. Recent work has demonstrated that the spine characteristics (vertebrae number, vertebrae length), of the thoracolumbar region, can be reliably measured from CT scans. Using such method, it was also identified that these characteristics exhibit significant intra-breed variation in Texel sheep, for example, the number of thoracolumbar vertebrae varied between 17 and 21 (Donaldson et al., 2013). Therefore, using CT measured spine traits this study investigates if any association exists between the TM-QTL genotype, or its pattern of inheritance, and spine characteristics.

2. Materials and methods

2.1. Animals sampled

Choosing only animals with an unambiguous TM-QTL genotype, the present study used a subset (n = 142) of the purebred Texel lamb records previously used by Macfarlane et al. (2010) and Lambe et al. (2011). Lambs were sired by seven different rams that were previously identified as carriers of at least one copy of TM-QTL. In order to classify the lambs TM-QTL genotype (homozygote non-carrier, +^S/+^D; heterozygote carrier inheriting TM-QTL from the sire, TM^S/+^D; heterozygote carrier inheriting TM-QTL from the dam, +^S/TM^D; homozygote carrier, TM^S/TM^D) all lambs were blood-sampled soon after birth (born 2009). Detailed information on the genotyping of the animals can be found in Macfarlane et al. (2010). The 142 lamb records used in this study divided into the TM-QTL genotype groups as follows: 39 +^S/+^D, 52 TM^S/+^D, 17 +^S/TM^D, 34 TM^S/TM^D and included 59 entire males and 83 female lambs from the purebred population of Texel sheep kept across two sites, one in Scotland and one in Wales, which were reared as either singles (n = 97), twins (n = 34), or artificially (pet; n = 11) (further details on the management of these animals can be found in Macfarlane et al. (2010) and Lambe et al. (2011)).

2.2. Computed tomography (CT) measurements

Lambs were CT scanned at ∼126 days of age (ranging from 93 to 145 days) and their topogram images (produced from the CT process) used to quantify spine characteristics for each, details provided in Donaldson et al. (2013). In short, spine traits measured directly from the scans included counts of vertebrae in the thoracic and lumbar regions (VN_THOR and VN_LUM respectively) and length (mm) of the thoracic and lumbar spine region (SPL_THOR and SPL_LUM respectively). These measures were used to calculate the average length (mm) of individual vertebrae in the thoracic and lumbar regions (VL_THOR (SPL_THOR/VN_THOR) and VL_LUM (SPL_LUM/VN_LUM) respectively). The results for the thoracic and lumbar spine regions were further used to provide the number of thoracolumbar vertebrae (VN_T+L (VN_THOR + VN_LUM)), and the length (mm) of the thoracolumbar region (SPL_T+L (SPL_THOR + SPL_LUM)). These thoracolumbar spine traits were then used to calculate the average length (mm) of individual vertebrae across the thoracolumbar region (VL_T+L (SPL_T+L/VN_T+L)).

For each lamb, the dimensions, width (mm), depth (mm), and area (mm²), of the longissimus lumborum (CT_MLL_W, CT_MLL_D, and CT_MLL_A, respectively), were estimated (from cross-sectional CT scans taken at the fifth lumbar vertebra) and given by Macfarlane et al. (2010) and included in this study's analysis of the genotypic effect. Essentially, these traits were included to, (i) determine if analysis of the reduced sample of animals shows genotype effects on loin traits similar to that observed for the larger sample, and, (ii) assess, from further analysis of the smaller data set, if the same or similar pattern of TM expression (polar overdominance) can be considered as a source for any genotype differences.

2.3. Statistical analysis

Data were analysed using the GLM procedure in SAS (SAS Institute Inc., Cary, NC, USA) to determine the effects of genotype on the collated loin traits and measured spine traits. Fixed effects fitted in the model for loin dimensions (CT_MLL_W, CT_MLL_D, CT_MLL_A), spine length traits (SPL_THOR, SPL_LUM, SPL_T+L, VL_THOR, VL_LUM, VL_T+L) and spine count traits (VN_THOR, VN_LUM, VN_T+L) were site, with two levels (Scotland and Wales), sex, with two levels (male and female), rearing rank, with three levels (single, twin, or pet), and TM-QTL-genotype, with four levels (+^S/+^D, TM^S/+^D, +^S/TM^D and TM^S/TM^D). The model was run with and without covariate adjustments for live weight (LWT); where any of the traits differed significantly between genotype groups, it was of interest to assess if, by testing the groups at a standard live weight, the differences were removed.

A set of orthogonal contrasts, as described by Freking et al. (1998a), (additive (−1 0 0 1), dominance (−1 1 1 −1), and reciprocal heterozygote (0 1 −1 0)) was fitted to the +^S/+^D, TM^S/+^D, +^S/TM^D, TM^S/TM^D genotypes, respectively. The contrasts test for any distinct pattern in the differences amongst the genotype group's least-squares means (for loin and spine traits), from which, a particular model for TM gene action might be suggested. Due to the previous evidence supporting the expression of the TM muscling phenotype through a polar overdominant mode of inheritance (Macfarlane et al., 2010), if significant differences were found between the heterozygote groups (reciprocal heterozygote test) a further set of orthogonal contrasts was fitted to the genotypes to include a test for the paternally derived polar overdominant mode of inheritance. Again following the approach from Freking et al. (1998a), this second set of orthogonal contrasts included additive (−1 0 0 1), maternal dominance (−1 0 2 −1), and polar overdominance (−1 3 −1 −1) models of gene action, which were fitted to the +^S/+^D, TM^S/+^D, +^S/TM^D, TM^S/TM^D genotypes, respectively. Contrasts were performed on the spine count data which was not adjusted for LWT and on the loin muscle and spine length data after the adjustment for LWT.

3. Results

In the context of this work it is useful to note that an earlier study (Macfarlane et al., 2012) found that least-squares means for LWT (measured at birth, 5, 10, 15, and 20 weeks of age) and carcass weight for TM^S/TM^D animals were consistently larger than that measured for +^S/+^D, +^S/TM^D, and TM^S/+^D genotype lambs (these differences were significant between TM^S/TM^D and +^S/+^D lambs for LWT at birth, 5, and 10 weeks of age, and carcass weight). In the present study, statistical models were initially run without an adjustment of LWT but given the above, statistical models were run again with certain traits (loin dimensions, spine length) adjusted for LWT to remove, as far as possible, any misinterpretation of TM-QTL effects. The following sections discussing these traits will therefore focus only on the LWT adjusted results.

3.1. Loin dimensions

Similar to the findings of Macfarlane et al. (2010), TM^S/+^D genotype lambs were observed to have the largest loin width, depth, and area, on average (Table 1). The differences in loin dimensions between TM^S/+^D and +^S/+^D genotype lambs were consistently significant, however, the larger trait averages observed for the TM^S/+^D group were not all significantly different from those averages observed for the +^S/TM^D and TM^S/TM^D genotype groups. For example, the TM^S/+^D group was significantly different from the +^S/TM^D group for loin area (CT_MLL_A_LWT) but the groups were not significantly different for loin width and depth measures (CT_MLL_W_LWT and CT_MLL_D_LWT, respectively). Further to this, and in contrast to Macfarlane et al. (2010), the TM^S/+^D group in this smaller data set was not significantly different from TM^S/TM^D in regards to all three loin dimension traits.

Table 1.

Least-squares means (and standard errors) for live weight, loin dimensions^b and spine traits^c for Texel lambs of each TM-QTL genotype.^d

Trait	Genotype								Site	Sex	Rearing rank	TM-QTL genotype	Live weight	R2a
	+S/+D		TMS/+D		+S/TMD		TMS/TMD
	n = 39		n = 52		n = 17		n = 34
Live weight	30.53	(1.004)	31.78	(0.855)	30.59	(1.402)	31.45	(1.094)	<0.001	0.104	<0.001	0.672	–	0.253
CT_MLL_W	66.27b	(1.034)	69.49a	(0.881)	67.12a,b	(1.444)	68.82a	(1.127)	0.084	0.069	0.199	0.034	–	0.126
CT_MLL_D	28.45b	(0.779)	30.96a	(0.663)	29.28a,b	(1.087)	30.44a	(0.848)	0.234	0.411	0.001	0.029	–	0.163
CT_MLL_A	1684c	(64.94)	1883a	(55.33)	1689b,c	(90.68)	1851a,b	(70.76)	0.385	0.633	<0.001	0.021	–	0.157
CT_MLL_W_LWT	66.27b	(0.591)	68.43a	(0.507)	67.06a,b	(0.825)	68.04a	(0.646)	<0.001	0.400	0.059	0.001	<0.001	0.716
CT_MLL_D_LWT	28.45b	(0.560)	30.28a	(0.481)	29.25a,b	(0.782)	29.94a	(0.612)	0.007	0.008	0.148	0.029	<0.001	0.570
CT_MLL_A_LWT	1684c	(37.68)	1817a	(32.36)	1685b,c	(52.63)	1803a,b	(41.17)	<0.001	0.002	0.353	0.004	<0.001	0.718
VNTHOR	12.69b,c	(0.064)	12.65c	(0.055)	12.89a,b	(0.090)	12.88a	(0.070)	0.026	0.350	0.044	0.006	.	0.174
VNLUM	6.356a	(0.074)	6.387a	(0.063)	6.111b	(0.104)	6.143b	(0.081)	0.092	0.427	0.061	0.009	.	0.149
VNT + L	19.05	(0.063)	19.04	(0.054)	19.00	(0.088)	19.02	(0.069)	0.759	0.981	0.645	0.967	.	0.011
SPLTHOR	255.6	(3.941)	260.7	(3.358)	257.9	(5.504)	262.5	(4.295)	0.003	0.429	0.034	0.503	.	0.122
SPLLUM	184.4a,b	(2.411)	186.8a	(2.054)	176.6c	(3.366)	180.3b,c	(2.627)	0.208	0.742	0.029	0.018	.	0.125
SPLT + L	440.0	(5.138)	447.5	(4.378)	434.5	(7.176)	442.8	(5.599)	0.004	0.447	0.015	0.329	.	0.128
VLTHOR	20.13	(0.276)	20.60	(0.235)	20.00	(0.386)	20.38	(0.301)	0.009	0.246	0.007	0.352	.	0.133
VLLUM	29.03	(0.292)	29.29	(0.249)	28.91	(0.408)	29.36	(0.319)	<0.001	0.702	0.045	0.657	.	0.124
VLT + L	23.10	(0.265)	23.51	(0.226)	22.86	(0.370)	23.28	(0.289)	0.004	0.443	0.007	0.332	.	0.135
SPLTHOR_LWT	255.6	(2.495)	256.9	(2.143)	257.7	(3.485)	259.7	(2.726)	0.379	0.456	0.088	0.610	<0.001	0.650
SPLLUM_LWT	184.4a	(2.033)	185.2a	(1.746)	176.5b	(2.839)	179.1b	(2.221)	0.151	0.513	0.333	0.007	<0.001	0.382
SPLT + L_LWT	440.0	(2.720)	442.0	(2.336)	434.2	(3.798)	438.8	(2.971)	0.061	0.241	0.575	0.289	<0.001	0.758
VLTHOR_LWT	20.13	(0.160)	20.32	(0.138)	19.99	(0.224)	20.18	(0.175)	0.051	0.772	0.595	0.531	<0.001	0.710
VLLUM_LWT	29.03	(0.147)	28.97	(0.126)	28.89	(0.205)	29.13	(0.160)	0.350	<0.001	0.011	0.721	<0.001	0.781
VLT + L_LWT	23.10	(0.130)	23.22	(0.112)	22.85	(0.182)	23.07	(0.142)	0.023	0.185	0.488	0.297	<0.001	0.793

Within a row, means with common letters (a–c), or no letters, in their superscript are not significantly different (P > 0.05).

LWT, live weight (kg) fitted as a covariate in model.

^aProportion of variance accounted for by the whole GLM model.

^bCT_MLL_W, width of M. longissimus lumborum (mm); CT_MLL_D, depth of M. longissimus lumborum (mm); CT_MLL_A, area of M. longissimus lumborum (mm²).

^cVN_THOR, number of thoracic vertebrae; VN_LUM, number of lumbar vertebrae; VN_T+L, number of thoracolumbar vertebrae; SPL_THOR, length of thoracic spine region (mm); SPL_LUM, length of lumbar spine region (mm); SPL_T+L, length of thoracolumbar spine region (mm); VL_THOR, average length of individual thoracic vertebrae (mm); VL_LUM, average length of individual lumbar vertebrae (mm); VL_T+L, average length of individual thoracolumbar vertebrae (mm).

^d+^S/+^D, homozygote non-carrier; TM^S/+^D, heterozygote carrier inheriting TM-QTL from sire; +^S/TM^D, heterozygote carrier inheriting TM-QTL from dam; TM^S/TM^D, homozygote carrier.

In general, the pattern of results from the analysis of the full data set (Macfarlane et al., 2010) suggested that the effect of the TM allele on these loin dimensions is expressed through a non-additive mode of inheritance (paternal polar overdominance). Analysis of the subset of records used here suggests a more general paternal TM-QTL effect on the loin with little evidence of a polar overdominance effect.

3.2. Spine length traits

Overall, there was no significant effect of the TM-QTL on the thoracic region length traits (SPL_THOR, VL_THOR). Nor was there an effect of TM-QTL genotype on the average length of individual lumbar vertebrae (VL_LUM), however, associations were shown to exist between TM-QTL genotype groups and length of the lumbar region (SPL_LUM) (Table 1). Least-squares means showed that, on average, +^S/+^D and TM^S/+^D genotype lambs had a longer lumbar length compared to +^S/TM^D and TM^S/TM^D genotype lambs. However, when considering the combined length of the thoracic and lumbar regions (SPL_T+L), the genotype effect is negligible (Table 1).

3.3. Spine count traits

The segmentation and anatomical regionalisation of the spinal elements (vertebrae) is established in early development (Wellik, 2007; Iimura et al., 2009), hence, it should not be affected by varying LWT. In running the statistical model with the inclusion and omission of a covariate adjustment of LWT, little difference was found between the least-squares means for each model (Table 1), lending support to the previous statement. Therefore, only the results obtained from the model without LWT adjustment will be discussed (results from the model with LWT covariate adjustment are not shown).

Regarding vertebrae number in the separate thoracic and lumbar spine regions first (VN_THOR and VN_LUM respectively), there were some significant differences between the genotype groups, however, the magnitude of these differences was relatively small (Table 1). In more detail, it can be seen from the least-squares means that there is much overlap between the genotype classes with regards to VN_THOR. The +^S/TM^D and TM^S/TM^D genotype lambs had, on average, a greater number of thoracic vertebrae than +^S/+^D and TM^S/+^D genotype lambs, however +^S/TM^D and +^S/+^D genotype lambs were not significantly different from each other. With regards to VN_LUM, the +^S/+^D and TM^S/+^D genotype lambs were significantly different from the +^S/TM^D and TM^S/TM^D genotype lambs. While observed to possess fewer thoracic vertebrae, +^S/+^D and TM^S/+^D genotype lambs had a greater number of lumbar vertebrae, on average.

Although significant differences occurred between the genotype groups for the two spine regions when considered separately, when examining the results for the combined thoracic and lumbar vertebrae number (VN_T+L), there were no significant differences between the groups (Table 1).

3.4. Orthogonal contrasts

Previous work on loin dimensions had shown strong evidence that the mode of inheritance for the TM-QTL deviates from a simple additive model (Macfarlane et al., 2010). Although the results obtained in this study's subset of data did not fully provide the same results, there was certainly an indication for superior loin dimensions in TM-QTL carrier lambs, especially in those with a paternal copy of the TM-QTL. Due to this, sets of orthogonal contrasts were fitted to the genotypes to investigate the situation further. These contrasts allowed testing for any particular patterns in the differences among the TM-QTL genotype (least-squares) means, for loin and spine traits, in order to define if certain modes of gene action may be present.

The first set of orthogonal contrasts was fitted to the genotypes to test for additive, dominance, and reciprocal heterozygote models of gene action (Table 2; only traits where TM-QTL genotype had a significant effect are shown).

Table 2.

Estimates of TM-QTL genotype^a contrasts (and standard errors) and significance levels (P-value) for additive, dominance, reciprocal, and paternally derived polar overdominant effects on loin dimensions^b and spine traits.^c

Trait	Additive		Dominance		Reciprocal heterozygote		Polar overdominance
	Contrast	P-value	Contrast	P-value	Contrast	P-value	Contrast	P-value
CT_MLL_W_LWT	1.767 (0.736)	<0.05	1.184 (1.151)	0.31	1.368 (0.893)	0.13	3.919 (1.698)	<0.05
CT_MLL_D_LWT	1.496 (0.697)	<0.05	1.142 (1.090)	0.30	1.032 (0.847)	0.22	3.207 (1.609)	<0.05
CT_MLL_A_LWT	118.9 (46.92)	<0.05	15.90 (73.38)	0.83	131.5 (56.98)	<0.05	279.0 (108.3)	<0.05
VNTHOR	0.186 (0.080)	<0.05	−0.032 (0.126)	0.80	−0.243 (0.097)	<0.05	−0.518 (0.185)	<0.01
VNLUM	−0.213 (0.092)	<0.05	−0.001 (0.144)	0.99	0.275 (0.112)	<0.05	0.550 (0.212)	<0.05
SPLLUM_LWT	−5.321 (2.531)	<0.05	−1.857 (3.958)	0.64	8.675 (3.073)	<0.01	15.49 (5.840)	<0.01

LWT, live weight (kg) fitted as a covariate in model.

^a+^S/+^D, homozygote non-carrier; TM^S/+^D, heterozygote carrier inheriting TM-QTL from sire; +^S/TM^D, heterozygote carrier inheriting TM-QTL from dam; TM^S/TM^D, homozygote carrier.

^bCT_MLL_W, width of M. longissimus lumborum (mm); CT_MLL_D, depth of M. longissimus lumborum (mm); CT_MLL_A, area of M. longissimus lumborum (mm²).

^cVN_THOR, number of thoracic vertebrae; VN_LUM, number of lumbar vertebrae; VN_T+L, number of thoracolumbar vertebrae; SPL_THOR, length of thoracic spine region (mm); SPL_LUM, length of lumbar spine region (mm); SPL_T+L, length of thoracolumbar spin region (mm); VL_THOR, average length of individual thoracic vertebrae (mm); VL_LUM, average length of individual lumbar vertebrae (mm); VL_T+L, average length of individual thoracolumbar vertebrae (mm).

The additive inheritance model was fitted as −1 0 0 1 to the +^S/+^D, TM^S/+^D, +^S/TM^D, and TM^S/TM^D genotypes respectively; testing the difference between the means of the homozygote genotypes. Where the contrast value was positive this showed that TM^S/TM^D had a larger mean than +^S/+^D for that particular trait and vice versa if the contrast value was negative. The difference between +^S/+^D and TM^S/TM^D genotype means was significant for all three loin traits CT_MLL_W_LWT, CT_MLL_D_LWT, and CT_MLL_A_LWT, and spine traits, VN_THOR, VN_LUM, and SPL_LUM_LWT. The dominance inheritance model was fitted as −1 1 1 −1 to the +^S/+^D, TM^S/+^D, +^S/TM^D, and TM^S/TM^D genotypes respectively; testing the combined means of the heterozygote genotypes (TM^S/+^D, +^S/TM^D) with the combined means of the homozygote genotypes (+^S/+^D, TM^S/TM^D). However, none of the differences between genotype means were significant, providing no evidence of a dominance effect on any of the traits. The reciprocal heterozygote model of gene action was fitted as 0 1 −1 0 to the +^S/+^D, TM^S/+^D, +^S/TM^D, and TM^S/TM^D genotypes respectively. This contrast tested the difference between the means of the two heterozygote genotypes (TM^S/+^D and +^S/TM^D), which were significant for traits CT_MLL_A_LWT, VN_THOR, VN_LUM, and SPL_LUM_LWT.

Freking et al. (1998a) previously commented that in such a case where the reciprocal heterozygote contrast is shown to be significant, the dominance contrast may be misleading i.e. under and over-estimation of heterozygote genotypes, and further analysis required. Therefore, due to this, and with the previous observation of the TM allele's expression through a non-additive mode of inheritance, a further set of orthogonal contrasts, additive, maternal dominance, and polar overdominance, were fitted to the genotypes as (1 0 0 −1) (−1 0 2 −1) (−1 3 −1 −1), respectively. Results for the additive model have been discussed above, and with no significant results for a maternal dominance effect only the results for the polar overdominance model from this set of contrasts were shown (Table 2) and discussed further.

The polar overdominance inheritance model was fitted as −1 3 −1 −1 to the +^S/+^D, TM^S/+^D, +^S/TM^D, and TM^S/TM^D genotypes respectively and used to test the difference between the mean of the TM^S/+^D group with each of the means calculated for +^S/+^D, +^S/TM^D, and TM^S/TM^D genotype groups. Contrast values for the paternal polar overdominance model are the combined differences between genotype means (condition as defined above) and were shown to be significant for all traits tested (Table 2); with the exception of VN_THOR, TM^S/+^D genotype lambs had a larger mean compared with each of the other genotype groups.

4. Discussion

It should be noted that the data set used in the present study was limited in its size, largely due to the restricted availability of sires (identified as TM-QTL carriers) which could be used to produce a study group of lambs. Nonetheless, to date, it is the only available data set which provides detail of the TM-QTL status for a sufficient number of purebred Texel animals, from which, the effects of TM-QTL on carcass, meat quality, and production traits could be assessed.

The analysis in the present report made use of lamb records, where TM-QTL genotype was unambiguously known, to, (i) determine if similar conclusions for loin dimensions could be formulated using only a subset of data in the analysis, repeating, as close as possible, the model described by Macfarlane et al. (2010), (ii) extend this test to determine if there is an effect of TM-QTL on underlying spine characteristics as the loin muscle is located parallel to spinal vertebrae, and, (iii) fit sets of contrasts to the TM-QTL genotype groups in order to determine the inheritance pattern of the TM-QTL.

It should also be noted that the following discussion will continue to refer only to loin dimension and spine length trait results generated from the model where all lamb records were adjusted for LWT.

4.1. TM-QTL and loin dimensions

Regarding the loin dimensions, CT_MLL_W, CT_MLL_D, and CT_MLL_A, the least-squares means for these traits reported by Macfarlane et al. (2010) are in strong agreement with an overdominance mode of expression of the TM allele; there is evidence of both heterozygote groups lying outside, in this case above, the phenotypic range of the homozygote groups. The results of Macfarlane et al. (2010) even suggested, more specifically, a paternally expressed polar overdominance effect as TM^S/+^D genotype lambs consistently exceeded +^S/TM^D genotype lambs in trait means; the difference between heterozygote groups, however, only appeared to be significant for the loin area (CT_MLL_A).

From the present study, the polar overdominance test (Table 2) did show significance, but the pattern of differences between the TM^S/+^D genotype least-squares means and the least-squares means for +^S/+^D, +^S/TM^D, and TM^S/TM^D genotype groups (Table 1) conflicts with this outcome and could not support a polar overdominance mode of TM gene action i.e. trait means for TM^S/+^D genotype lambs did not appear to significantly ‘out-perform’ over all (or the majority) of the other genotype groups (Table 1). Nonetheless, the pattern of least-squares means did still infer TM-QTL expression which could not be explained by simple additive gene action, and though an overdominance model could not be supported, there was still indication towards some paternal influence of the TM allele; genotype groups which inherited a copy of the TM allele from the sire (TM^S/+^D and TM^S/TM^D) were observed to have, on average, larger loin width (CT_MLL_W_LWT), depth (CT_MLL_D_LWT), and area (CT_MLL_A_LWT) measures.

4.2. TM-QTL and spine characteristics

Three out of the nine spine traits (VN_THOR, VN_LUM, and SPL_LUM) were observed to be significantly different amongst the TM-QTL genotype groups (Table 1).

The least-squares means for vertebrae number traits (VN_THOR, VN_LUM) followed a curious pattern; means for +^S/+^D and TM^S/+^D genotype lambs were similar and significantly different, in most instances, to the means for +^S/TM^D and TM^S/TM^D genotype lambs, the latter of which were also similar to each other in their mean values. The +^S/+^D and TM^S/+^D groups had, on average, fewer thoracic vertebrae (VN_THOR) but more lumbar vertebrae (VN_LUM), with the situation reversed for the +^S/TM^D and TM^S/TM^D groups. However, the overall number of thoracolumbar vertebrae (VN_T+L) across the four genotype groups was not different. It appeared that TM-QTL inheritance patterns had no substantial effect on the total number of thoracolumbar vertebrae but may have some influence on the thoracic-lumbar vertebral arrangement in the spine, but this remains uncertain as the size of difference between genotype groups is small.

Similarly, the total length of the thoracolumbar region (SPL_T+L) was not significantly different between the genotype groups. Freking et al. (1998b) observed shorter spinal columns in CPLG genotype lambs, and given that TM-QTL falls close to its position on the chromosome, it was an important point to investigate further in connection with TM-QTL inheritance. Least-squares means for spine length traits (SPL, VL) from the present study, however, do not suggest any such negative effects of TM-QTL on spine length. The TM^S/+^D genotype lambs, which express the muscle hypertrophy phenotype, in fact, were observed to have, on average, longer thoracolumbar vertebrae (VL_T+L_LWT), and subsequently longer thoracolumbar spine regions (SPL_T+L_LWT), but these trait values (23.22 mm; 442.0 mm respectively) were not significantly different from the other three groups. It is interesting that, in the animals with the largest loin dimension measures (TM^S/+^D), there appeared to be no substantial changes to the skeletal structure on which this muscle lies.

Given that differences in spine traits are largely non-significant between the genotype groups and that the overall pattern of least-squares means is indistinct, the models of gene action should be interpreted with caution. For example, the contrast tests showed significance for the polar overdominant model of TM-QTL gene action on VN_THOR, VN_LUM, and SPL_LUM spine traits. These results should, again, be carefully considered alongside least-squares means (Table 1) as, though slightly larger (for VN_LUM and SPL_LUM), the means for TM^S/+^D genotype lambs did not significantly ‘out-perform’ over all other genotype groups for these spine traits. Hence, there was no strong indication that the observed differences in spine trait phenotypes were associated with increased loin muscling specific to TM-QTL gene action.

What seems important to take from the present study is that increased loin muscling, particularly associated with TM^S/+^D genotype lambs, has been shown to have little association with the underlying spine characteristics. Information on spine characteristics, in general, could potentially be used to improve loin production, i.e. through increasing the size and/or number of loin chops (Donaldson et al., 2013). Hence, it would be interesting to investigate further the potential size of increase in loin production from those TM^S/+^D animals which possess a greater number of thoracolumbar vertebrae.

5. Conclusion

Given the results from the present study, it was evident that some effect of the TM-QTL on loin dimension phenotypes was linked to a paternal genetic influence, but, with a weaker data set (67 fewer records) this study could not provide further evidence for a specific polar overdominance inheritance pattern. With regards to spine characteristics, in general terms, the analysis of the subset of data did not reveal any obvious (advantageous or disadvantageous) associations with TM-QTL inheritance. There did not appear to be any effect on spine/vertebrae length and detailing how, or if, the TM-QTL interacts in the vertebral patterning process (given the thoracic-lumbar vertebral combinations across genotype groups) would require analysis of a substantially larger data set than what was available at present.