Parameters and functional analysis of the deep epaxial muscles in the thoracic, lumbar and sacral regions of the equine spine

Abstract

The epaxial muscles produce intervertebral rotation in the transverse, vertical and axial axes. These muscles also counteract the movements induced by gravitational and inertial forces and movements produced by antagonistic muscles and the intrinsic muscles of the pelvic limb. Their fascicles are innervated by the dorsal branch of the spinal nerve, which corresponds to the metamere of its cranial insertion in the spinous process. The structure allows the function of the muscles to be predicted: those with long and parallel fibres have a shortening function, whereas the muscles with short and oblique fibres have an antigravity action. In the horse, the multifidus muscle of the thoracolumbar region extends in multiple segments of two to eight vertebral motion segments (VMS). Functionally, the multifidus muscle is considered a spine stabiliser, maintaining VMS neutrality during spine rotations. However, there is evidence of the structural and functional heterogeneity of the equine thoracolumbar multifidus muscle, depending on the VMS considered, related to the complex control of the required neuromuscular activity. Osteoarticular lesions of the spine have been directly related to asymmetries of the multifidus muscle. The lateral (LDSM) and medial (MDSM) dorsal sacrocaudal muscles may be included in the multifidus complex, the function of which is also unclear in the lumbosacral region. The functional parameters of maximum force (F _max), maximum velocity of contraction (V _max) and joint moment (M) of the multifidus muscles inserted in the 4th, 9th, 12th and 17th thoracic and 3rd and 4th lumbar vertebrae of six horses were studied postmortem (for example: 4MT4 indicates the multifidus muscle that crosses four metameres with cranial insertion in the T4 vertebra). Furthermore, the structural and functional characteristics of LDSM and MDSM were determined. Data were analysed by analysis of variance (anova) in a randomised complete block design (P ≤ 0.05). For some muscles, the ordering of V _max values was almost opposite to that of F _max values, generally indicating antigravity or dynamic functions, depending on the muscle and VMS. The muscles 3MT12, 3ML3 and 4ML4 exhibited high F _max and low V _max values, indicating a stabilising action. The very long 7MT4 and 8MT4 multifidus had low F _max and high V _max values, suggesting a shortening action. However, some functional characteristics of interest did not fall within these general observations, also indicating a dual action. In summary, the results of the analysis of various structural and functional parameters confirm the structural and functional heterogeneity of the equine thoracolumbar multifidus complex, depending on the VMS, regardless of the number of metameres crossing each fascicle. To clarify the functions of the equine multifidus muscle complex, this study aimed to assess its functional parameters in thoracolumbar VMSs with different movement characteristics and in the MDSM and LDSM muscles, hypothesising that the functional parameters vary significantly when the VMS is considered.

Introduction

The epaxial muscles are defined by their innervation by dorsal branches of the spinal nerves occupying the vertebral groove. Their anatomic position enables intervertebral rotational movements along three axes: transverse, vertical and axial (Barone, 1989; Haussler, 1999; Groesel et al. 2010). Their counteraction of the passive movements induced by gravitational and inertial forces and the active movements produced by antagonistic and intrinsic muscles of the pelvic limbs are also important, as Schilling & Carrier (2009) have shown in dogs.

In the thoracolumbar region, the epaxial muscles consist of long, superficial fascicles forming the erector spinae muscles (iliocostalis, longissimus and spinalis), whereas the semispinalis muscle with long fascicles is not differentiated in mammals because it is incorporated into the spinae muscle, according to Barone (1989). Also included in the transverse spinae muscle is the multifidus muscle, which has deep fascicles extended caudally from lumbar mammillary processes, thoracic transverse processes and cervical articular procesess to the lamina and the lateral side of the spinous processes. Stubbs et al. (2006) include the lateral (LDSM) and medial (MDSM) dorsal sacrocaudal muscles within the multifidus complex.

The metamere is an embryological concept and constitutes the segment extending between the cranial and caudal halves of two contiguous vertebrae, including the bony and muscular planes and the corresponding spinal nerve (Barone, 1989). The multi‐segmented multifidus muscle crosses a number of different metameres and, in this case, its fascicles are innervated by the dorsal branch of the spinal nerve that corresponds to the metamere of its cranial insertion into the spinous process (Macintosh et al. 1986).

The multifidus muscle is considered a spine stabiliser in pigs (Kaigle et al. 1995), humans (Macintosh et al. 1986) and horses (van Weeren et al. 2010), avoiding an abnormal rotation during the contraction of antagonist muscles such as the abdominal oblique muscles. In an evolutionary study of small marsupials and placental mammals, Schilling (2009) reports that the components of the multifidus muscle complex have both stabilising and dynamic functions.

In the equine thoracolumbar region, there is evidence of variation in the number of metameres crossing the multifidus muscles. Stubbs et al. (2006) and Hyytiäinen et al. (2014) indicate multifidi with one to five metameres. Barone (1989) argues that multifidi cross two or three metameres, considering the 2‐metamere multifidi to be a long rotator muscle (R). Mc Gowan et al. (2007) report the existence of multifidi with one to four metameres, and Nickel et al. (1985) cite multifidi extending between twio and six metameres. In a recent study, García Liñeiro et al. (2017) show the presence of multifidus muscles in the equine thoracolumbar region occupying between two and eight metameres, with the longest located in the interscapular region, whereas only two to four metameres occur from the 12th thoracic vertebra towards caudal.

This multi‐segment structure of the multifidus muscle functionally means that it crosses one or more vertebral motion segments (VMS). The VMS anatomically consists of adjacent vertebrae and interposed soft tissue structures (Haussler, 1999), forming the spinal unit of movement. During rotational movements in the three axes of VMS, axial rotation, lateral blending and flexion‐extension, the cranial vertebra moves over a fixed point (caudal vertebra). During flexion, the centre of rotation (CR) is located in the centre of the caudal vertebral body, which moves dorsally during extension and ventrally and cranially toward the intervertebral disc during flexion (Denoix, 1999).

Evidence in humans, pigs and horses on the stabilising function of the VMS suggests that the deep epaxial muscles maintain VMS neutrality during extension activities of the superficial epaxial muscles (longissimus and spinalis) and during flexion activities of hypaxial muscles (psoas and abdominal muscles; van Weeren, 2009). Stubbs et al. (2011) report the existence of equine osteoarticular diseases and thoracolumbar pain directly related to asymmetries and atrophies of the multifidus muscle. Stubbs et al. (2006), Mc Gowan et al. (2007), Stubbs et al. (2011), van Weeren et al. (2010) and Clayton et al. (2012) have reported the importance of the multifidus muscle in horses as a VMS stabiliser. Hyytiäinen et al. (2014) argue that, during equine locomotion, spinal function requires complex control of the neuromuscular activity in the thoracolumbar and lumbosacral regions. Given this complexity, García Liñeiro et al. (2017) suggest that the equine multifidus muscle in horses is an immunohistochemically homogeneous muscle with various architectural designs that have functional significance according to the VMSs. These structural and functional differences are independent of the predominance of type I and IIA fibres within this muscle complex, according to Hyytiäinen et al. (2014) and García Liñeiro et al. (2017). The structure of these multi‐segment muscles described by García Liñeiro et al. (2017), with short fibres of different lengths firmly held in aponeurotic sheets, are optimal structures for movement control in each VMS, according to Higham & Biewener (2011). The presence of short fibres in the structure of a muscle provides the advantage of mechanically adequate activity with low energy consumption when the function of a muscle is to maintain tendon tension (Ker et al. 1988).

To clarify the functions of the equine multifidus muscle complex, this study aimed to assess its functional parameters (F _max, V _max and M) in thoracolumbar VMSs with different movement characteristics and in the MDSM and LDSM muscles, hypothesising that the functional parameters vary significantly when the VMS is considered.

Materials and methods

This study continues research already published by García Liñeiro et al. (2017) using the same material and the same values of muscle mass, normalised sarcomere fibre length and pennation angle of muscle fibres within the muscle. Briefly, six Silla Argentino horse cadavers were used, with ages ranging from 11 to 15 years, which arrived at the Department of Anatomy of the School of Veterinary Sciences, University of Buenos Aires (Universidad de Buenos Aires – UBA) within 24 h of their death and with no history of locomotor injuries. The thoracolumbar column was collected by performing cross‐sections at the 6th cervical vertebra cranially and at the height of the bodies of both ilium bones caudally. After removing the erector spinae, preserving the septum and separating it from the multifidus muscle and disarticulating the sacroiliac joints, paramedian sections were made through a plane across the neck of the ribs and lumbar transverse processes. The sections were submerged for 72 h in a 10% formalin solution to preserve the physiological length of the muscles during fixation, in accordance with Lieber & Fridén (2000). Subsequently, on the right side, a careful dissection of the deep spinal transverse muscle was performed in each of the VMSs of interest, from the cranial insertion in the spinous process of thoracic vertebrae 4, 9, 12 and 17 and lumbar vertebra 3 (T4, T9, T12, T17 and L3) to their caudal insertion adjacent to the transverse or mammillary processes, in accordance with Stubbs et al. (2006). The muscles studied were named according to the number of metameres they cross and their vertebra of cranial insertion (for example: 4MT4 indicates the multifidus muscle that crosses four metameres with cranial insertion in the T4 vertebra); the muscles that cross two metameres were considered to be long rotators (R) (Barone, 1989). The LDSM muscles from their cranial insertion in the spinous process (L4–L5) to their caudal insertion in the 4th caudal vertebra and the MDSM muscles from their cranial insertion in the spinous process in the 3rd sacral vertebra (S3) to their caudal insertion shared with the LDSM were removed. The muscle fascicles thus collected were placed in a 10% formalin solution in water for 7 days, followed by a drip rinse with running water (6 h) and then immersed in a 15% nitric acid solution for 3 days, according to a methodological adaptation by Hermanson (1997), Delp et al. (2001) and Sharir et al. (2006), taking the dimensions of the muscles into account. When the fascicles could be separated with blunt instruments, the muscles were again rinsed in running water (24 h), immersed in 0.4 m phosphate‐buffered saline (PBS) solution for 2 h and finally placed in a 6% formalin solution at 4 °C until analysis. To study their architecture, the muscles were removed from the formalin solution, rinsed for 10 min in running water and dried at room temperature for 30 min for uniform dehydration. The length of the treated fascicles was measured using a Vernier scale (accurate to 0.01 mm), after which a blunt dissection of muscle fascicles was performed using an illuminated stand magnifier and identifying the parallel aponeurotic sheets in which the fascicles are inserted. The length between the aponeurotic sheets of fascicles insertion was measured using a scale; then, the lengths of at least 10 fascicles per muscle were measured. To determine a precise measure of fibre length, the values obtained using the scale were normalised (NFL) to sarcomere lengths using the equation FL × OSL/MSL, in accordance with Felder et al. (2005), where FL is the length measured using a scale, OSL is the equine optimal sarcomere length (2.80 μm) according to Marx et al. (2006) and MSL is the measured value of the mean sarcomere length. The MSL value was calculated after measuring the sarcomere lengths of 10 fibres from each studied muscle in 1.000×, 150‐pixel digital images in a TIFF format using the open‐access image analyser scion image 4.0, β version. The length between aponeurotic sheets and fibre length data were converted to a 10× scale. On graph paper, the aponeurotic sheets were drawn as two parallel lines, placing a scale line between both representing the length of the muscular fascicle; then, the angle formed between the lines representing the fascicle and the aponeurotic sheets was measured using a goniometer, indirectly determining the pennation angle of muscle insertion (ɵ). After recording these data, the fibre muscles of the fibrous connective component that form the tendon/aponeurotic sheets, the epimysium and peripheral walls were dissected in each muscle using an illuminated stand magnifier, a left‐hand clamp and a scalpel. The connective and muscle tissues were weighed separately on a precision balance (≥ 0.001 g) to obtain the connective tissue mass and muscle mass (Mm). The effective tendon length (l) was measured in MDSM and LDSM muscles, according to the equation by Ker et al. (1988), I = D‐NFL, where D is the tendon length from the origin of insertion.

To assess the moment arm (r) of the muscles of interest, one of the preparations was placed on a flat surface and digital photographs were taken with a camera located laterally in the same dorsal plane as the vertebrae, carefully determined using a level. Images were recorded serially, in a caudal to cranial direction. The focus of each image was centred on the muscles inserted in each VMS of interest. Subsequently, using digital photographs in TIFF format with 150‐pixel resolution and the open‐access image analyser scion, β version, the line of action of the muscle of interest was drawn, from which a perpendicular line that joins this line with the centre of rotation of the vertebra during VMS extension (Denoix, 1999) was drawn, and r was measured according to Williams et al. (2008). Using this datum, the moment of force (M) on the joint of each muscle was calculated according to the equation proposed by Williams et al. (2008): M = F _max × r.

The following functional parameters were assessed using the data collected: F _max, expressed in Newtons (N): [(Mm × cos ɵ)/ρ × NFL] × 22.5 Ncm⁻², where Mm is expressed in grams, ρ is the muscle density (1.056 g cm⁻³, Lieber & Ward, 2011), NFL is expressed in cm, and the factor 22.5 N cm⁻² indicates the maximum isometric stress of the vertebrate skeletal muscle (Lieber & Ward, 2011). The maximum contraction velocity (V _max) was expressed in m s⁻¹: (5) × (type I RA) × (0.33) × (NFL/s) + (5) × (type IIA RA) × (1.33) × (NFL/s) + (5) × (type IIX RA) × (3.20) × (NFL/s). NFL is expressed in metres. In the equation constructed to calculate V _max, the factors 0.33, 1.33 and 3.20 correspond to the maximum contraction velocity of types I, IIA and IIX equine skeletal muscle fibres, respectively, in vitro, at 15 °C, according to Rome et al. (1990); the multiplication by 5 is a correction factor for the contraction velocity of muscle fibres at equine body temperature, according to Payne et al. (2005). RA is the relative area of each type of muscle fibre studied, according to García Liñeiro et al. (2017). A sample of the centre of the muscle was collected from the left side of each anatomical piece, covered with talcum powder, frozen by immersion in liquid nitrogen, and stored at −80 °C until analysis to determine the RA of each type of fibre in the LDSM and MDSM muscles. The activity of the myofibrillar enzyme myosin adenosine triphosphatase (mATPase) was assessed in 9‐μm‐thick serial sections of each fibre type upon acidic pre‐incubation (pH 4.3) and using monoclonal antibodies specific for equine heavy‐chain myosin isoforms, A.474 (positive for isoform IIA and intermediate for isoform IIX) and BA‐F8 (positive for isoform I), following the immunohistochemical procedure described by García Liñeiro et al. (2017). After identifying the fibres of the acidic pre‐incubation mATPase reactions in TIFF digital images at 150‐pixel resolution using the software motic plus 2.0, the cross‐sectional area (CSA) of each fibre was determined and, from these data, the RA occupied by each type of fibre was assessed using the open‐access image analyser scion image, β version.

Statistical analysis

Values of F _max, V _max and M were analysed by an anova using a randomised complete block design in which the animals were considered the blocks (n = 6; P ≤ 0.05). The significance of any differences was analysed using the Bonferroni test as post hoc analysis. In all cases, normality of the values was tested using the Shapiro–Wilk test with a 10% significance level. To reach an acceptable power level (Ghasemi & Zahediasl, 2012), the significance level of the Shapiro–Wilk test was tested at 10% to minimise the probability of type II errors. Only parametric tests were performed because normality was observed in the analysis of all variables.

For further details on materials, methods and muscle structure please refer to García Liñeiro et al. (2017).

Results

The F _max, V _max and M values are shown in Tables 1, 2 and 3 respectively. The values of the ratio between NFL and r (NFL/r) increased (but not significantly) and are shown in Fig. 1. The functional ordering (stabilisation vs. shortening) of the muscles studied is shown in Fig. 2. MDSM and LDSM insert in the caudal vertebrae, which are clearly distinguished from the first five caudal vertebrae. MDSM fibres were oriented in a lateral to medial direction, constituting a voluminous muscle inserted in the spinous process S3. The LDSM, located lateral to the MDSM, originates from small cylindrical tendons in the caudal column oriented laterally from the first caudal segments, constituting a voluminous muscle, which extends to segment S4, and continuing cranially with a long tendon that is inserted in the lumbar spinous processes L5–L4. The lengths of the belly and tendon of the LDMS muscle are similar (Fig. 3). The mean numerical values for the MDSM and LDSM muscle structures are outlined in Table 4.

Table 1.

Average values of maximum force (F _max) indicated in N (Newtons) of each of the muscles studied

Multifidus	F max (N)	SE
8MT4	22.93b	37.1
3MT4	50.83b	25.9
RL3	62.25b	25.9
RT4	68.68b	25.9
RT9	88.63b	25.9
7MT4	94.0b	28.5
6MT4	101.39b	25.9
RT17	103.73b	25.9
RT12	109.06b	25.9
4MT4	124.51b	28.5
3MT9	128.20b	25.9
4MT9	140.50ab	25.9
MDSM	143.10ab	25.9
LDSM	147.88ab	25.9
4MT12	156.74ab	25.9
4MT17	160.76ab	25.9
4ML3	163.49ab	25.9
3MT17	172.30ab	25.9
3MT12	222.38ab	25.9
3ML3	276.64a	25.9

SE, standard error.

Identical superscripts indicate lack of significance (P ≤ 0.05).

Table 2.

Mean values of V _max for each muscle

Muscle	V max	SE
RL3	0.03c	0.01
RT9	0.03c	0.01
RT17	0.04c	0.01
3ML3	0.04c	0.01
4ML3	0.04c	0.01
4MT4	0.04c	0.01
3MT12	0.04c	0.01
RT4	0.04c	0.01
RT12	0.04bc	0.01
6MT4	0.05bc	0.01
3MT17	0.05bc	0.01
3MT9	0.06bc	0.01
MDSM	0.07bc	0.01
4MT12	0.07bc	0.01
3MT4	0.08bc	0.01
4MT9	0.08bc	0.01
8MT4	0.09bc	0.01
7MT4	0.09bc	0.01
4MT17	0.11b	0.01
LDSM	0.30a	0.01

SE, standard error.

Values expressed in m s⁻¹. Identical superscripts indicate lack of significance (P ≤ 0.05).

Table 3.

Values of muscle moment (r) and joint moment (M) exerted by each muscle

Muscle	r (cm)	M (N cm−1)	Rank	SE
RL3	6.5	414.72	4.05c	2.3
8MT4	15.5	514.36	4.08bc	3.4
3MT4	9.5	489.40	4.83bc	2.3
RT4	7.5	510.12	5.15bc	2.3
RT9	6	610.471	5.32bc	2.3
RT17	6	638.02	6.22bc	2.3
RT12	6.6	774.44	7.20bc	2.3
3MT9	8.5	1176.02	10.90bc	2.3
3MT17	7.5	1243.09	12.92abc	2.3
7MT4	15.5	1229.64	13.93abc	2.6
4MT4	11.5	1368.81	14.32abc	2.6
4MT17	9	1569.99	14.47abc	2.3
4ML3	10	1884,01	14.63abc	2.3
6MT4	15	1664,34	15.21abc	2.3
4MT9	11.5	1699.26	16.16abc	2.3
4MT12	11	1368.81	17.24ab	2.3
3MT12	8.5	2069.34	18.90ab	2.3
3ML3	8.5	2481.35	23.51a	2.3

SE, standard error.

Significant values of M were analysed as rank. Identical superscripts indicate lack of significance (P ≤ 0.5).

Figure 1.

Relation between LFN (cm) and r (cm) in each muscle.

Figure 2.

Pairing between F _max and V _max values. In each muscle, the greater amplitude between both parameters indicates higher differentiation between shortening capacity vs. antigravity function.

Figure 3.

(A) 1, the LDSM are shown with the tendon inserted cranially in the vertebrae L5‐L4; 2, MDSM. The markers are inserted in the spinous processes from cranial to caudal corresponding to the V (5) and VI (6) lumbar vertebrae, I (1) and V (5) sacral vertebrae and I (1) caudal vertebra. (B) The caudal tendons of the LDSM (1) and MDSM (2) are inversely oriented.

Table 4.

Numerical values of the MDSM and LDSM muscle structures

Muscle	Insertions	Metameres	NFL (cm)	SE	Mm	SE	D (cm)	l (cm)
LDSM	C5‐L5	11	4.25a	0.18	106.83a	6.66	66	61.75
MDSM	C5‐S3	7	3.19b	0.14	105.33a	6.66	24	20.81

The NFL and Mm values with the same superscripts indicate lack of significance (p > 0.05).

D, tendon length; l, effective tendon length; Mm, muscle mass; NFL, normalised fibre length; SE, standard error.

Discussion

The findings of this study analysing various structural and functional parameters confirm the structural and functional heterogeneity of the equine thoracolumbar multifidus complex, as previously suggested by García Liñeiro et al. (2017), depending on the VMS and regardless of the number of metameres crossing each muscle.

Data on muscle architecture were collected by studying preserved cadavers. An alternative would have been to perform imaging studies, such as computed tomography and magnetic resonance imaging (Thompson et al. 2011). Lieber & Fridén (2000), Delp et al. (2001) and Sharir et al. (2006) consider the study of muscle architecture using preserved cadavers the most suitable method. They argue that muscle architecture cannot be determined by magnetic resonance, computed tomography or ultrasound because these imaging studies do not measure fibre length or variations in the insertion angle of the fibres that occur along the muscle, particularly in muscles with complex structures.

Payne et al. (2005) assume that type IIA fibres are predominant in equine muscles and calculate the V _max using the value of the maximum shortening velocity of this type of fibre (1.33 m s⁻¹), according to Rome et al. (1990). In the present study, to increase the accuracy of the V _max calculation, the RA of each muscle fibre type composing each muscle was considered, in accordance with a previous study by García Liñeiro et al. (2017).

The results indicate that, for some muscles, the V _max values were ordered almost opposite to F _max values (as shown in Tables 1 and 2). In general, this indicates antigravity or dynamic functionality, depending on the muscle and VMS, corroborating the established general concept that muscles are functionally classified as generators of isometric force or producers of rapid shortening (Burkholder et al. 1994; Kearns et al. 2002; Felder et al. 2005; Payne et al. 2005; Eng et al. 2008). This shows that the NFL is directly proportional to the capacity to generate V _max and inversely proportional to the capacity to generate F _max. The pairings between F _max and V _max values (Table 5, Fig. 2) indicate that the greater the difference between values, the greater the polarisation between the shortening or stabilisation functions of each muscle.

Table 5.

Congruence (mating) between the numerical values representing the increasing order of the values of V _max and F _max of the muscles studied

Muscle	F max	V max
8MT4	1	7
3MT4	2	6
RL3	3	1
RT4	4	2
RT9	5	1
7MT4	6	7
6MT4	7	3
RT17	8	2
RT12	9	2
4MT4	10	2
3MT9	11	4
4MT9	12	6
MDSM	13	5
LDSM	14	9
4MT12	15	5
4MT17	16	8
4ML3	17	2
3MT17	18	3
3MT12	19	2
3ML3	20	2

See Tables 1 and 2. At each pairing, the greater amplitude between both parameters indicates greater differentiation of function between shortening contraction vs. isometric contraction.

In some muscles, a more detailed analysis of both functional parameters, F _max and V _max, shows functional characteristics of interest beyond these general concepts.

The 4M muscles of the T17, T9 and T12 VMSs have the highest range of significant V _max values (Table 2), and 4MT17 has the highest significant value. This homogeneous ordering of the 4M muscles in the analysed VMS would suggest an increased shortening action in the transversal (VMS T17), vertical and axial (VMS T9‐T14) axes during rotation, according to Townsend & Leach (1984), Denoix (1999), Licka et al. (2009) and Groesel et al. (2010). However, their F _max values (Table 1) are also among the highest, with equal significance, very similar absolute values, and high M values (Table 3). These findings suggest a dual action in these muscles of both antigravity and shortening, which is highly selective in the studied VMSs (Fig. 2), indicating heterogeneity, specificity and complexity in the innervation and function of the equine thoracolumbar multifidus musculature. Furthermore, within this dual behaviour, the value of the NFL/r ratio (Fig. 1) indicates that the dynamic capacity is ordered as T17 > T9 > T12, considering that the NFL/r ratio is directly proportional to the capacity of a muscle to rotate a joint across wide ranges (Williams et al. 2008). According to Schilling (2009), the dual behaviour of these 4M muscles would represent the global, dynamic and stabilising epaxial muscles cited in an evolutionary study in small terrestrial mammals.

The 3M muscles of the same VMS mentioned in the previous paragraph (T9, T12 and T17) have medium V _max values, which are similar in T9 and T17, and a low V _max value in T12 (T9 > T17 > T12), whereas the F _max values are significantly ordered as T12 > T17 > T9, indicating a very selective, dual, heterogeneous, stabilising and shortening action in the T9 and T17 VMS, in agreement with the increases of lateroflexion in T9 and flexion in T17 (Denoix, 1999) and differing from the clearly stabilising action with predominantly isometric contraction (Felder et al. 2005) in the T12 VMS (Table 5 Fig. 2). The M and NFL/r values of the 3MT9 and 3MT17 are similar (Fig. 1), showing the same capacity to rotate the joint, according to Williams et al. (2008). The values of the same parameters of the 3MT12 indicate that this muscle acts as an isometric force producer and is an agonist of the longissimus muscle, which acts as a spine stabiliser. This muscle has greater and preventive activity of the moment arm of the ipsilateral pelvic limbs, with the highest activity occurring in T12, according to Licka et al. (2009) and Groesel et al. (2010).

In the lumbar region, the muscles inserted in the L3, 3ML3 and 4ML3 VMSs have very low values of V _max and low NFL/r ratios, and very high values of F _max and M, clearly indicating a stabilising action in the lumbosacral column, which has little rotation in the L2–L5 segment, albeit supporting the very high transverse rotation (extension) at the lumbosacral level and the sacral axial rotation, according to Townsend & Leach (1984), Denoix (1999) and Zaneb et al. (2013). Figure 2 shows that the 3ML3 and 4ML3 muscles have similar amplitudes, indicating a homogeneous stabilising function. This stabilising action of the lumbar multifidus muscles consistently contributes to the lumbar muscle platform formed by the longissimus muscle, as suggested by Robert et al. (2001) and Licka et al. (2009), together with the likely stabilising action of the MDSM (Nickel et al. 1985; Hyytiäinen et al. 2014).

MDSM and LDSM have become muscles of interest in recent years. The dual actions of some of the studied muscles suggested in this study have also been proposed by Hyytiäinen et al. (2014) for the LDSM, based on its high percentage of IIA fibres. This indicates functional complexity and the need for neuromuscular coordination of this muscle complex. Stubbs et al. (2006) argue that the MDSM and LDSM set correspond to the multifidus complex and has an unclear function in the lumbosacral region. We consider the LDSM to be the longissimus, whereas the MDSM corresponds to the multifidus muscle, based on fibre orientation (Fig. 3), fibre type composition and in accordance with Nickel et al. (1985) and Barone (1989). Based on the value of the LDSM tendon length and its cranial position relative to the muscle, we find its function as an elevator and lateral flexor of the tail unlikely, at least exclusively, as proposed by Getty et al. (1975), Barone (1989) and Evans (1993). Therefore, we propose that considering its structure, the action of this muscle contributes to the lumbosacral extension, at least at some phase of movement. The structure of these muscles, with short fibres and long tendons (61.75 cm for the LDSM and 20.81 cm for the MDSM), could be a mechanically suitable muscle‐tendon system, with the benefits of a low activation energy to maintain tension with a small muscle mass, according to Ker et al. (1988); theoretically, these muscles may act by muscle fibre stretching, elongating the tendon in the same direction (exerting negative work and receiving energy from the external system), by short contraction (exerting positive work with energy released to the tendon and the environment) or by both mechanisms of fibre stretching and contraction. Horses develop these muscular‐skeletal designs in a repeated and extreme way, as Butcher et al. (2009) report for the equine superficial digital flexor muscle, a structure of short fibres specialised in economically generating increased strength and storing elastic energy.

Considering the values of V _max and F _max assessed, both SCD muscles have an amplitude value (Fig. 2) that provides them with a dual function, with a higher stabilising capacity in the MDSM and a higher joint rotation capacity in the LDSM.

The multifidus muscles 7M and 8M inserted in the VMS T4 along with the 3MT4 have very low and significant values of F _max, suggesting that they could act as agonists of the longissimus and spinalis muscles cranially to the thoracolumbar platform, according to García Liñeiro et al. (2017). In the thoracolumbar platform, the longissimus muscle acts as a powerful spine stabiliser, as reported by Robert et al. (2001) and Licka et al. (2009), transmitting propulsion impulses from the pelvic limbs. The significantly high values of V _max of these muscles confirm their dynamic capacity, particularly of the 7MT4 muscle, and high rotation capacity, based on their values of M (Tables 3 and 5) and NFL/r ratios, contributing to the extension of the interscapular region (withers; García Liñeiro et al. 2017) and opposing the flexion during the descent of the neck (Denoix, 1999). The high RA values in the long muscles 7MT4 and 8MT4 (García Liñeiro et al. 2017) of type I fibres, which have the lowest shortening velocity according to Rome et al. (1990), albeit the most efficient low‐velocity movements (Bottinelli & Reggiani, 2000), are compatible with their high V _max values because the architectural design of these muscles shows that they have fibres with long lengths normalised to the length of the sarcomere, according to García Liñeiro et al. (2017). This architectural design is suitable for shortening and is in accordance with the extensive evidence presented by Burkholder et al. (1994), Lieber & Fridén (2000), Kearns et al. (2002), Felder et al. (2005) and Eng et al. (2008), who consider that the architectural design, not the fibre type composition, of a muscle is the main predictor of its function.

Moseley et al. (2002) report that the human rotator muscles have very low moment arm values because their insertions are very close to the centre of rotation of the vertebrae, opposing movement by compression. The same anatomical characteristic is found in horses (García Liñeiro et al. 2017), and the present findings indicate very low M values (Table 3).

We are aware that one of the weaknesses of this study is that the capacity of movement was calculated based on the capacity of fibre contraction, that is, considering positive work based on muscle structural characteristics. In turn, the existence of negative work, accumulating energy in the tendon through an elongation mechanism of the tendon‐fibre assembly, should not be overlooked (Ker et al. 1988; Butcher et al. 2009).

We agree with Hyytiäinen et al. (2014), who argue that the concept of Moseley et al. (2002), that those deep fascicles of the multifidus control inter‐segmental movement and more superficial fascicles control spinal orientation, cannot be fully transposed to equines.

Electromyography is a useful tool with which to study muscle activity during movement. However, this resource requires quantitative data on the muscle‐tendon assembly, including length, maximum contraction velocity and force generated, to understand fully the role of the muscle, according to Harrison et al. (2012). Accordingly, we believe that the present study may be useful as a basis for morphological, functional and applied studies with the aim of understanding equine spine pathologies and their treatments based on previously proposed mobilisation and dynamic posture exercises (Clayton et al. 2012).