Adaptive functional specialisation of architectural design and fibre type characteristics in agonist shoulder flexor muscles of the llama, Lama glama

Authors

  • Guillermo H. Graziotti,

    1. Departament of Veterinary Anatomy, Faculty of Veterinary Sciences, University of Buenos Aires, Buenos Aires, Argentina
    Search for more papers by this author
  • Verónica E. Chamizo,

    1. Laboratory of Muscular Biopathology, Department of Comparative Anatomy and Pathological Anatomy, Faculty of Veterinary Sciences, University of Cordoba, Cordoba, Spain
    Search for more papers by this author
  • Clara Ríos,

    1. Departament of Veterinary Anatomy, Faculty of Veterinary Sciences, University of Buenos Aires, Buenos Aires, Argentina
    Search for more papers by this author
  • Luz M. Acevedo,

    1. Laboratory of Muscular Biopathology, Department of Comparative Anatomy and Pathological Anatomy, Faculty of Veterinary Sciences, University of Cordoba, Cordoba, Spain
    Search for more papers by this author
  • J. M. Rodríguez-Menéndez,

    1. Departament of Veterinary Anatomy, Faculty of Veterinary Sciences, University of Buenos Aires, Buenos Aires, Argentina
    Search for more papers by this author
  • C. Victorica,

    1. Departament of Veterinary Anatomy, Faculty of Veterinary Sciences, University of Buenos Aires, Buenos Aires, Argentina
    Search for more papers by this author
  • José-Luis L. Rivero

    1. Laboratory of Muscular Biopathology, Department of Comparative Anatomy and Pathological Anatomy, Faculty of Veterinary Sciences, University of Cordoba, Cordoba, Spain
    Search for more papers by this author

Dr José-Luis L. Rivero, Laboratory of Muscular Biopathology, University of Cordoba, Faculty of Veterinary Sciences, Campus de Rabanales, Edificio de Sanidad Animal, Crtra. Madrid a Cádiz Km 396, 14014 Córdoba, Spain. E-mail: an1lorij@uco.es

Abstract

Like other camelids, llamas (Lama glama) have the natural ability to pace (moving ipsilateral limbs in near synchronicity). But unlike the Old World camelids (bactrian and dromedary camels), they are well adapted for pacing at slower or moderate speeds in high-altitude habitats, having been described as good climbers and used as pack animals for centuries. In order to gain insight into skeletal muscle design and to ascertain its relationship with the llama’s characteristic locomotor behaviour, this study examined the correspondence between architecture and fibre types in two agonist muscles involved in shoulder flexion (M. teres major – TM and M. deltoideus, pars scapularis – DS and pars acromialis – DA). Architectural properties were found to be correlated with fibre-type characteristics both in DS (long fibres, low pinnation angle, fast-glycolytic fibre phenotype with abundant IIB fibres, small fibre size, reduced number of capillaries per fibre and low oxidative capacity) and in DA (short fibres, high pinnation angle, slow-oxidative fibre phenotype with numerous type I fibres, very sparse IIB fibres, and larger fibre size, abundant capillaries and high oxidative capacity). This correlation suggests a clear division of labour within the M. deltoideus of the llama, DS being involved in rapid flexion of the shoulder joint during the swing phase of the gait, and DA in joint stabilisation during the stance phase. However, the architectural design of the TM muscle (longer fibres and lower fibre pinnation angle) was not strictly matched with its fibre-type characteristics (very similar to those of the postural DA muscle). This unusual design suggests a dual function of the TM muscle both in active flexion of the shoulder and in passive support of the limb during the stance phase, pulling the forelimb to the trunk. This functional specialisation seems to be well suited to a quadruped species that needs to increase ipsilateral stability of the limb during the support phase of the pacing gait. Compared with other species, llama skeletal muscles are well suited for greater force generation combined with higher fatigue resistance during exercise. These characteristics are interpreted as being of high adaptive value, given the llama’s habitat and its use as a pack animal.

Introduction

Examination of the anatomical and physiological constraints that determine muscle function provides a useful insight into muscle design. Among functionally equivalent muscles, there may be substantial variation in activation, strain, architecture and fibre-type composition depending on the specific functional role of the muscle (see Higham & Biewener, 2011 for a recent review). Ultimately, muscle architecture and fibre-type composition are probably the two most important design parameters that determine the functional properties of a muscle (Roy et al. 1984; Burkholder et al. 1994; Eng et al. 2008). Muscle architecture is defined as the arrangement of muscle fibres relative to the axis of force generation (Lieber & Friden, 2000). Muscle force is primarily determined by the cross-sectional area (CSA) of the fibres (Powell et al. 1984), whereas muscle excursion and velocity are determined by muscle fibre length (Bodine et al. 1982). Thus, the architectural features of a muscle define its functional properties. In addition, muscle fibre type influences contractile force, maximum shortening velocity and resistance to fatigue (Bottinelli & Reggiani, 2000). Differences in contractile force between fibre types are not simply a function of differences in fibre CSA, they also result from the specific contractile properties of the myosin heavy chain (MyHC) isoforms they express (Bottinelli & Reggiani, 2000). Furthermore, there are relevant metabolic differences among the various fibre types (Peter et al. 1972); the fibre-type composition of a given muscle is thus closely aligned with its biochemical properties.

Although ample evidence exists for the impact of architecture and fibre type on muscle functional properties, the relationship between these two parameters has not yet been fully clarified. Because architectural properties and fibre types affect muscle function, information regarding their covariance is of value. Many studies have reported a clear correlation between muscle architectural design and fibre-type composition (Roy et al. 1984; Gellman et al. 2002; Graziotti et al. 2004). For example, muscles or muscle sub-volumes with long fibres and low pinnation angles tend to have a large number of fast-twitch fibres, small to moderate fibre sizes, weak oxidative capacity and few capillaries (Graziotti et al. 2004). This muscle design implies a relatively high percentage of half sarcomeres arranged in series (Roy & Ishihara, 1997), which would optimise displacement and velocity of shortening (Roy et al. 1984), suggesting significant excursions of these sub-volumes related to dynamic joint excursions. In contrast, the muscles or muscle sub-volumes packaged with shorter fibres and higher pinnation angles display a clear tendency towards slow-oxidative fibre-type features – a high proportion of type I fibres, large fibre sizes, high oxidative capacity and considerable capillarisation (Graziotti et al. 2004). This implies a relatively high proportion of half sarcomeres arranged in parallel (Roy & Ishihara, 1997), suggesting less potential for excursion changes, but a more important anti-gravitational postural role.

Few studies have examined a sufficient number of muscles to enable a clear link to be confirmed between differences in muscle design and muscle function. To our knowledge, the only two animals for which the entire hindlimb muscle architecture and fibre-type composition have simultaneously been measured are the mouse (Burkholder et al. 1994) and the rat (Eng et al. 2008). These laborious studies demonstrated that architectural properties were a greater predictor of muscle function (as defined by primary joint action and anti-gravity or non anti-gravity role) than fibre type. Architectural properties were not strictly correlated with fibre type, but when muscles were grouped according to their anti-gravity vs. non anti-gravity function there was evidence of functional specialisation. Anti-gravity muscles had a larger percentage of slow fibre types and greater muscle physiological cross-sectional area (PCSA) than non-anti-gravity muscles. The authors of these studies concluded that the lack of correlation between a muscle’s (expected) architecture and fibre type may reflect the variability of functional requirements in single muscles (Eng et al. 2008). An important weakness of these two studies (openly expressed by their authors) was that the very small size of both the mouse and the rat results in a fibre-type composition that is heavily biased toward fast fibres. This limits the value of regression analysis between muscle architecture and fibre-type composition, as only a relatively narrow range of one of the independent variables (i.e. fast fibre-type percentage) is available. In consequence, and even though the results of these studies are valuable for understanding how muscles are designed, full elucidation of the relationship between muscle architecture and fibre type proved impossible.

The llama (Lama glama) is one of the few mammals of relatively large body size in which the four main myosin fibre types (I, IIA, IIX and IIB) are extensively expressed in locomotor muscles (Graziotti et al. 2001). As the fibre-type composition of this species is clearly not biased toward either slow or fast phenotypes (Graziotti et al. 2004), it may provide a useful animal model for investigating the relationship between muscle architecture and fibre type.

Relatively little is known about the relationship between muscle features and the characteristic locomotor behaviour displayed by these animals in response to their habitat. The domestic llama is a member of the artyodactyl clade Camelidae, which comprises the New World (wild guanaco, vicuña, alpaca and llama) and Old World (bactrian and dromedary camels) subclades. They were domesticated in pre-Columbian South America about 6000–7000 years ago (Kadwell et al. 2001), and for a long time were used as pack animals at altitudes of over 4000 m a.s.l., occupying ecosystems harsher (colder and drier) than those inhabited by other domestic animals (e.g. alpacas and sheep; Iñiguez et al. 1998). As prolonged and active sojourn at high altitudes implies limited oxygen availability (Jurgens et al. 1988), one might reasonably expect llama muscles to display structural and functional adaptations similar to those observed after prolonged aerobic resistance training – a high proportion of slow fibre types and larger CSA of muscle cells compatible with an increase in muscle force-generating potential, and a parallel increase in fibre capillarisation and oxidative potential (MacDougall et al. 1991; Green et al. 1998; Holm et al. 2010; Doria et al. 2011).

Moreover, camelids are distinguished from most cursorial mammals by their natural ability to pace (moving ipsilateral limbs in near synchronicity), and indeed prefer this footfall pattern over others, particularly at slow to moderate speeds (Van der Sluijs et al. 2010; Pfau et al. 2011). It has been suggested that pacing evolved as a means of avoiding limb interference (Webb, 1972). However, lateral stability is often believed to be reduced, because the pacing animal shifts its weight from one side to the other, resulting in side-sway (Van der Sluijs et al. 2010). Nevertheless, all camelids seem morphologically well adapted to pacing (Janis et al. 2002): the chest and hips are very narrow for the overall size of the animal, and the dorsally tapering abdomen is narrower than the trunk of other ungulate species, such as cattle, deer or horses (Pfau et al. 2011). They also have relatively long limbs, allowing for a long stride (Pfau et al. 2011), and padded, digitigrade feet with two main supportive toes that might help with ipsilateral stability (Janis et al. 2002). As a result of all conformational and anatomical features, these animals have a very narrow-based stance, with the limbs very close to the midline, perhaps making them more stable when supported only by ipsilateral limbs (Webb, 1972).

Intriguingly, remarkable differences in locomotor behaviour have been reported between New and Old World camelids (Van der Sluijs et al. 2010; Pfau et al. 2011). Unlike the dromedary, alpacas and llamas cannot perform a symmetrical running gait, involving aerial phases when no limbs are on the ground (Van der Sluijs et al. 2010; Pfau et al. 2011). At faster speeds, they switch to asymmetrical gaits, predominantly transverse gallops, but they never pace or trot (Pfau et al. 2011). This characteristic locomotor behaviour of the New World camelids has been ascribed to a complex and fascinating evolution (Van der Sluijs et al. 2010; Pfau et al. 2011). In contrast to the desertic (or semi-desertic) habitats of the dromedary, which has to cover large distances in search of food and water, the habitat of the New World camelids provides a fairly good distribution of food and water; the mountainous Andean regions, for example, are characterised by considerable ecological differences, including deep valleys and high-altitude plains (altiplanos). Here, walking is the preferred gait among free-ranging animals, and galloping is rare. The llama’s preference for the walking gait can also be interpreted as a high-altitude adaptation, as very fast motion also implies higher demand for oxygen, which is limited at high altitudes (Jurgens et al. 1988). Thus, it may be postulated that the running gait used by Old World camelids would not be of great adaptive value for New World camelids in a mountainous habitat, or for animals that, unlike dromedaries, have been described as good climbers (Iñiguez et al. 1998; Van der Sluijs et al. 2010).

This study focused on two agonist muscles of the llama that are primarily flexors of the shoulder joint (i.e. M. teres major – TM and M. deltoideus, pars scapularis – DS and pars acromialis – DA). To the authors’ knowledge, no previously-published studies have addressed the morphological and functional organisation of these muscles in ungulate species. Previous architectural studies in the hare (Lepus europeus) and the racing greyhound (Canis familiaris) have shown that TM and DS muscles are characterised by long fibres and small PCSAs with respect to DA, reflecting adaptations for dynamic movements over a larger range of joint displacements in TM and DS muscles than in the DA muscle, predominantly used in anti-gravity or postural roles (Williams et al. 2007, 2008). Although TM and DS muscles are described anatomically as shoulder flexors (Dyce et al. 2010), electromyographic (EMG) research suggests additional functional roles of these two muscles, possibly linked to different topographical locations and activation patterns. Thus, the TM muscle can be considered both anatomically (Herring et al. 1993) and functionally (Hyland & Jordan, 1997) as a monoarticular head of the M. latissimus dorsi, that when the limb is in an outstretched position can also contribute simultaneously to medial movement (adduction) of the arm (Hyland & Jordan, 1997; Edge-Hughes, 2004). EMG activity of the TM muscle has also been reported in the support phase of the gait in dogs (Tokuriki, 1973) and cats (English, 1978), indicating that this muscle can also contribute to joint stability as an anti-gravitational muscle. The DS muscle is also described as a flexor of the shoulder but, unlike the TM, it is also active in the rat during abduction and as the humerus rotates forward, suggesting that it might also be involved in extension of the shoulder (Hyland & Jordan, 1997), as demonstrated in humans (van der Helm, 1994).

The llama TM muscle displays other gross anatomical features that could be linked to the particular locomotor behaviour described above. Unlike in carnivores and primates, but as in horses, ruminants and pigs (Barone, 1989), the insertion of the llama TM muscle into the medial aspect of the humerus is clearly more distal than the insertion of the DS muscle into the lateral aspect of the humerus (G. Graziotti, personal observation). This design suggests considerable leverage of the TM in the llama, providing a wide range of motion to the shoulder joint when the limbs swing forwards and backwards. Nevertheless, as the llama needs to increase ipsilateral stability during the support phase of its natural pacing gait, the TM muscle should be well adapted to contribute to the adduction of the limb during support, requiring force production and fatigue-resistance properties. It may therefore be postulated that the TM muscle of the llama might well be, in architectural terms, a fast muscle (long fibres, low PCSA) acting as a dynamic muscle in shoulder flexion during the swing phase of the gait, but equipped with a slow-oxidative fibre phenotype (i.e. a high proportion of type I fibres, large fibre sizes, abundant capillaries and high oxidative potential) for the continuous adduction of the limb during the support phase of the gait. This combination may be not only a very useful high-altitude adaptation, but also an optimal design for improving overall stability when the animal is supported only by its ipsilateral limbs.

The aim of the present study, therefore, was to examine from a comparative perspective the architectural features (i.e. fibre length, pinnation angle and PCSA) and fibre-type characteristics (i.e. proportion, size, capillarisation and oxidative potential of immunohistochemical muscle fibre types) in the agonist TM, DS and DA muscles of the llama. The objective was to contribute further to our understanding of muscle design in this species, and to gain insight into the relationship between muscle features and the characteristic locomotor behaviour of this New World camelid as part of its adaptation to the habitat.

Materials and methods

Ten clinically healthy adult (2.5–3 years old) llamas, of both genders (five males, five females), and of comparable size and conformation were used. All animals were slaughtered at the usual commercial body weight of 120 kg by exsanguination in compliance with Argentine rules on commercial slaughtering. The TM, DS and DA muscles from paired forelimbs were excised from carcasses within 1 h of death. Left muscles were used for morphological descriptions of muscle architecture, whilst the contralateral right muscles were used for analysis of fibre-type features.

Muscles for architectural design were cleaned of fat and connective tissue, and immersed in 10% buffered formalin for about 4 weeks (Ryan et al. 1992; Liu et al. 1997). Muscles were then placed in a 25% nitric acid solution for about 10 days (Graziotti et al. 2004). The duration of digestion was controlled until connective tissue attachments were comprised and individual muscle fibres bundles could be teased away from muscle sub-volumes. Muscles were then rinsed overnight in a 0.4 m phosphate-buffered solution at pH 7.2, and transferred to a 50% glycerol solution until soft and pliable. Several fibre bundles were dissected from different locations of each muscle sub-volume under a dissecting microscope, and fibre-bundle length and pinnation angle were measured; shortening or elongation of muscle tissue due to chemical fixation was deemed likely to be < 5% of in vivo bundle lengths (Roy et al. 1984). This technique yielded intact muscle fibre bundles in relatively natural positions, and enabled digestion of most of the connective tissue investing the muscles (Ryan et al. 1992). From these architectural data, PCSA for each muscle or muscle portion was calculated following Sacks & Roy (1982), using the method outlined by Graziotti et al. (2004).

Fresh muscle samples were collected from the core of each muscle or muscle sub-volume of right limbs within 3 h of slaughtering. In order to minimise the impact of fibre-type regionalisation, which frequently occurs within muscles (Kernell, 1998; Graziotti et al. 2004; Myatt et al. 2011), care was taken to standardise sampling sites within each muscle as accurately as possible with regard to relative depth and location along the medial–lateral and proximal–distal axes of muscles (Fig. 1). Thus, muscles were removed from their points of origin to the point of insertion, and sampling sites were fixed midway between these two points. In both TM and DA muscles, these sites corresponded to their proximal sub-volumes, as described in the gross dissection (see Results). Samples (about 1 cm3), oriented such that myofibres could be cut transversely, were frozen by immersion in isopentane kept in liquid nitrogen, and stored at −80 °C until analysed. Samples were serially cross-sectioned at 10 μm thickness in a cryostat microtome at −22 °C, and mounted on poly-l-lysine-coated glass slides for immunohistochemical and histochemical examination.

Figure 1.

 Blunt dissection of the left TM (A, lateral aspect), and DS and DA (B, medial aspect) of the llama after treatment with a 25% nitric acid solution (see Materials and methods for details) showing the architectural design of the muscles (explained in Results). Dr and Cd, dorsal and caudal borders of the muscles, respectively; A and It, aponeurosis and insertion tendons of the muscle, respectively. *The asterisk marks the position of a fibrous septum dividing the acromial part of the M. deltoideus into two sub-volumes, one proximal and another distal. p and d, denote proximal and distal sub-volumes, respectively, within TM and the acromial part of the M. deltoideus. X denotes sampling sites used for fibre typing in each muscle sub-volume.

Serial sections were reacted with a panel of MyHC isoform-specific monoclonal antibodies (Table 1), as previously described (Graziotti et al. 2004). An additional serial section reacted with the monoclonal antibody GSL-Isolectin B4 (Vector Laboratories, Burlingame, CA, USA) was used for visualising capillaries (Graziotti et al. 2004). Two more serial sections were stained for qualitative myofibrillar ATPase histochemistry after acid (pH 4.45, 2 min) and alkaline (pH 10.5, 10 min) preincubations, as detailed in Graziotti et al. (2004). A further serial section was also stained for nicotinamide adenine dinucleotide tetrazolium reductase (NADH-TR) according to Dubowitz & Sewry (2007), and used for quantitative assessment of oxidative capacity of single muscle fibre types as previously described (Graziotti et al. 2004). Numerical distribution of fibre types in each muscle sample was established by immunohistochemistry, as detailed in Graziotti et al. (2004). Moreover, for each individual fibre analysed (150–250 fibres per block sample), CSA and mean optical densities were measured (in duplicate) in NADH-TR histochemical reactions, and the absolute number of capillaries in contact with individual fibres was estimated in the section reacted with the antibody GSL I-Isolectin B4. Quantitative information (area, NADH-TR activity and capillaries) was averaged according to MyHC fibre type. To calculate the relative CSA occupied by each main fibre type (i.e. fibre-type composition in area), the seven immunohistochemically delineated fibre types (see Results) were collapsed into the four major types (I, IIA, IIX and IIB) by combining each half of these mixed fibre types with their respective pure fibre types. For example, half of the hybrid IIAX fibres were combined with type IIA fibres, and the other half with IIX fibres. The four resulting major fibre types were then combined with the respective area measurements to yield fibre-type percent area, which has been shown to correlate with relative MyHC isoform percentage (Linnane et al. 1999).

Table 1.   Specificity* of monoclonal antibodies (MAbs) against adult rat skeletal MyHC isoforms used in the study, and immunohistochemical and enzyme histochemical characterisation of skeletal muscle fibre types in llama.
MAbsSpecificity for rat MyHC isoforms
IIIAIIXIIB
  1. *According to Schiaffino et al. (1989) except for MAbs A4-74 (Hughes et al. 1993) and S5-8H2 (Rivero et al. 1999).

  2. According to Graziotti et al. (2001, 2004). For immunohistochemistry, + and –, positive and negative reactions, respectively, for that specific MyHC isoform or fibre type with that MAb. For enzyme histochemistry, +++, ++, + and –, very strong, strong, intermediate and weak reactions, respectively, for that specific muscle fibre type; the symbol ↔ indicates an intermediate histochemical reaction for that hybrid fibre type inbetween their respective pure muscle fibre types.

  3. MyHC, myosin heavy chain; mATPase, myofibrillar adenosine triphosphatase; NADH-TR, nicotinamide adenine dinucleotide tetrazolium reductase.

BA-F8+
SC-75+++
A4-74++
BF-35+++
S5-8H2+++
 Llama skeletal muscle fibre types, immunohistochemistry
MAbsII + IIAIIAIIAXIIXIIXBIIB
BA-F8++
SC-75++++++
A4-74+++++
BF-35++++
S5-8H2++++++
 Llama skeletal muscle fibre types, histochemistry
StainII + IIAIIAIIAXIIXIIXBIIB
mATPase, pH 4.45++++++
mATPase, pH 10.5++++++
NADH-TR++++++

Descriptive statistics were used to derive means ± SEM of muscle architectural data (muscle mass, fibre length, pinnation angle and PCSA) and muscle fibre-type characteristics (fibre-type composition, fibre CSA, mean number of capillaries in contact with fibre types and NADH-TR activity). Once Gaussian distribution was demonstrated, these variables were analysed using one-way analysis of variance. Significant differences among muscles or muscle sub-volumes were then determined using Tukey’s HSD post-hoc tests. Differences were considered significant at < 0.05.

Results

Muscle architecture

The mean values of architectural data for the three muscles or muscle sub-volumes examined in the llama are shown in Table 2. These data represent pooled means of six–eight measurements of fibre lengths and pinnation angles from a variety of regions within the muscle belly, but differences between regions within a muscle were not quantified. Variations of these muscular features between individuals were fairly low, as shown in Table 2.

Table 2.   Muscle mass, fibre length, pinnation angle and PCSA in agonist muscles (or muscle sub-volumes) examined in the llama.
Muscle featureTMDSDAP
  1. Values indicate mean ± SEM (n = 10), and P indicates P-values of the one-way analysis of variance carried out among muscles.

  2. Within a row, means with the same letter are not statistically different (> 0.05).

  3. DA, M. deltoideus, pars acromialis; DS, M. deltoideus, pars scapularis; PCSA, physiological cross-sectional area; TM, M. teres major.

Muscle mass (g)59.2 ± 6.2C51.2 ± 6.8B20.1 ± 0.9A< 0.001
Fibre length (mm)114.4 ± 2.3C85.0 ± 3.4B24.1 ± 1.7A< 0.001
Pinnation angle (°)8.0 ± 1.1A14.9 ± 1.0B35.0 ± 0.1C< 0.001
PCSA (mm2)48.5 ± 5.4B55.1 ± 8.5B65.5 ± 3.8A0.002

Gross dissection of the TM muscle was characterised by the presence of an aponeurosis on its caudomedial aspect (A in Fig. 1A). This aponeurosis consistently presented a cranial division reaching the distal (insertion) tendon of the muscle and a second caudal division blended with the lamina profunda of the fascia brachii. This muscle has the longest muscle fibres, with an average (combined) fibre length of 114.4 ± 2.3 mm, running in with very discrete pinnation angle (8.0 ± 1.1 °, range 6–10), from both the proximal (origin) tendon to the aponeurosis of the muscle (p in Fig. 1A), and from this aponeurosis to the distal tendon of the muscle (d in Fig. 1A). Overall, the combined PCSA of the TM muscle was 48.5 ± 5.4 mm2.

The DS muscle displayed a well-defined lateral aponeurosis arising from the caudal border and spine of the scapula, and ending on the middle third of the muscle belly (A in Fig. 1B). The muscle fibres of the scapular portion of the M. deltoideus, with an average length of 85.0 ± 3.4 mm and a low pinnation angle of 14.9 ± 1.0 °, were also arranged in parallel from different levels of the aponeurosis to the distal (insertion) tendon of the muscle (Fig. 1B). Nevertheless, these architectural data were significantly shorter and narrower, respectively, in DS than in the TM muscle (Table 2). However, the PCSA of this portion of the M. deltoideus (55.1 ± 8.5 mm2) was not significantly different to that of the TM muscle (> 0.05).

The short DA muscle was covered by a dense aponeurosis that extended both medially and laterally. This part of the muscle was divided into two sub-volumes, located proximal and distal to a grossly-visible fibrous septum on the medial aspect of the muscle (asterisk in Fig. 1B). Muscle fibres in the proximal sub-volume of this muscular partition were arranged diagonally between the lateral and medial aponeuroses of the muscle, with an average length of about 28 mm, and an average pinnation angle of about 30 ° (p in Fig. 1B). In the distal sub-volume of the acromial part of the M. deltoideus, located near to the distal (insertion) tendon of the muscle, myofibres were even shorter (about 20 mm on average), and had a greater pinnation angle (about 40 °) than in the proximal sub-volume (d in Fig. 1B). Combining both sub-volumes, the DA muscle had the shortest fibres (24.1 ± 1.7 mm), the highest pinnation angle (35.0 ± 0.1 °) and the largest combined PCSA (65.5 ± 3.8 mm2; Table 2).

Muscle fibre types

Seven fibre types were defined immunohistochemically as a function of MyHC content (Table 1). Four were pure fibre types expressing a single MyHC isoform (I, IIA, IIX and IIB), and three were hybrid phenotypes co-expressing two isoforms (I + IIA – result not shown, IIAX and IIXB; Fig. 3). The immunoreactivity of these fibre types with the battery of antibodies used here has previously been reported in llamas (Graziotti et al. 2001, 2004). On the basis of the myofibrillar ATPase reaction after acid preincubation at pH 4.45, muscle fibres were divided into four main categories, while hybrid fibres had intermediate reactions with respect to pure phenotypes (Table 1; Fig. 3a). The same was true for the myofibrillar ATPase histochemical reaction after alkaline preincubation at pH 10.5 (Table 1; result not shown in Fig. 3). According to NADH-TR staining, fibres were also qualitatively classified into different staining intensities with oxidative capacity decreasing in the rank order I > IIA > IIX > IIB (Table 1; Fig. 3b).

The three muscles or muscle sub-volumes examined in the present study differed very significantly in terms of percentages, CSA, capillaries and oxidative capacity of their various constituent fibre types (Tables 3 and 4). Both TM and DA muscles contained a significantly higher percentage (both in number and in area) of type I fibres, and a lower proportion of fast-glycolytic type IIB fibres than DS muscle. The percentage of IIB fibres was still significantly lower in DA than in TM. The relative frequency of type IIA fibres was significantly lower in DS than in DA; no statistically significant differences (> 0.05) were found between these two sub-volumes in the TM muscle. Type IIX muscle fibres did not vary significantly among muscles. Collectively, these data indicate that the I-to-II fibre ratio was significantly higher in both TM and DA muscles than in DS muscle, no significant difference being recorded between TM and DA (Fig. 2c, left vertical axis). The IIA-to-IIB fibre ratio was significantly higher in the DA muscle than in both TM (4.5-fold) and DS (fivefold) muscles, and also significantly higher in TM than in DS (Fig. 2c, right vertical axis).

Table 3.   Muscle fibre-type composition (%) in number and in relative area (%), and mean CSA (μm2) of fibre types in agonist muscles examined in the llama.
 TMDSDAP
  1. Values indicate mean ± SEM (n = 10), and P indicates P-values of the one-way analysis of variance carried out among muscles.

  2. *Hybrid fibres are considered with their respective pure fibre types (see Materials and methods). Within a row, means with the same letter are not statistically different (> 0.05). ND, not determined due to the low proportion of this particular fibre type.

  3. CSA, cross-sectional area; DA, M. deltoideus, pars acromialis; DS, M. deltoideus, pars scapularis; TM, M. teres major.

Fibre-type composition, in number (%)
 I38.9 ± 2.3B31.4 ± 2.1A39.9 ± 2.5B0.031
 IIA14.3 ± 2.1AB10.2 ± 1.1A19.0 ± 1.4B0.002
 IIAX3.5 ± 1.27.0 ± 3.25.1 ± 1.20.503
 IIX30.5 ± 1.129.1 ± 2.127.8 ± 2.50.636
 IIXB5.7 ± 1.44.7 ± 1.17.0 ± 2.20.638
 IIB7.5 ± 1.9B18.1 ± 3.9C1.1 ± 1.0A<0.001
Fibre-type composition, in area (%)*
 I36.9 ± 3.4B22.6 ± 2.6A36.4 ± 3.1B0.003
 IIA14.9 ± 1.913.5 ± 3.119.6 ± 1.50.167
 IIX37.1 ± 2.337.7 ± 4.042.5 ± 3.10.452
 IIB11.1 ± 2.1A26.2 ± 6.2B1.5 ± 1.3A<0.001
Mean CSA (μm2)
 I4901 ± 172B2479 ± 173A4845 ± 376B<0.001
 IIA5117 ± 204B3159 ± 268A4784 ± 248B<0.001
 IIAX6119 ± 3475214 ± 4446092 ± 3480.221
 IIX5522 ± 223B4509 ± 262A6333 ± 239B<0.001
 IIXB4889 ± 3895505 ± 1586255 ± 5670.065
 IIB6204 ± 2696537 ± 213ND0.077
 All5388 ± 156B4380 ± 147A5634 ± 278B<0.001
Table 4.   Mean number of capillaries in contact with each muscle fibre type (Cap/fib), and quantitative histochemical reaction of the NADH-TR activity for each fibre type (OD, optical densities) in agonist muscles examined in the llama.
 TMDSDAP
  1. Values indicate mean ± SEM (n = 10), and P indicates P-values of the one-way analysis of variance carried out among muscles.

  2. Within a row, means with the same letter are not statistically different (> 0.05).

  3. ND, not determined due to the low proportion of this particular fibre type.

  4. DA, M. deltoideus, pars acromialis; DS, M. deltoideus, pars scapularis; NADH-TR, nicotinamide adenine dinucleotide tetrazolium reductase; TM, M. teres major.

Mean absolute number of capillaries (cap/fib)
 I8.64 ± 0.16B7.50 ± 0.12A8.11 ± 0.35AB0.007
 IIA7.17 ± 0.16B6.25 ± 0.11A7.11 ± 0.28B0.004
 IIAX8.88 ± 0.25B6.84 ± 0.12A7.62 ± 0.31A< 0.001
 IIX7.53 ± 0.11B6.18 ± 0.07A7.12 ± 0.39B0.001
 IIXB6.35 ± 0.226.28 ± 0.096.41 ± 0.460.945
 IIB6.61 ± 0.086.60 ± 0.09ND0.901
 All7.53 ± 0.09B6.61 ± 0.05A7.30 ± 0.28B0.002
Histochemical reaction of NADH-TR (OD, optical densities)
 I0.63 ± 0.02AB0.60 ± 0.01A0.65 ± 0.01B0.003
 IIA0.38 ± 0.04AB0.35 ± 0.03A0.45 ± 0.02B0.002
 IIAX0.41 ± 0.06B0.22 ± 0.04A0.35 ± 0.08B< 0.001
 IIX0.27 ± 0.02B0.16 ± 0.01A0.26 ± 0.02B< 0.001
 IIXB0.21 ± 0.01B0.10 ± 0.01A0.25 ± 0.02B0.012
 IIB0.20 ± 0.02B0.09 ± 0.01AND0.006
 All0.42 ± 0.03B0.29 ± 0.01A0.45 ± 0.02B< 0.001
Figure 2.

 Two-dimensional relationship between architectural design parameters and fibre-type characteristics of the M. teres major (TM on the horizontal axis) and M. deltoideus, pars scapularis (DS) and pars acromialis (DA) of the llama (Lama glama). (a) Relationship between average muscle fibre length (left axis) and pinnation angle (right axis); p and d on the horizontal axis denote proximal and distal, respectively, sub-volumes within TM and the acromial part of the M. deltoideus, as explained in the Results. (Note that muscles or muscle sub-volumes with long fibres invariably had low pinnation angles and, conversely, short-fibred muscles had high pinnation angles.) (b) Relationship between physiological cross-sectional area (PCSA, left vertical axis) and average cross-sectional area (CSA) of individual muscle fibres. [Note how PCSA is directly proportional to fibre pinnation angle and inversely proportional to muscle fibre length – compare with (a), whereas CSA is not correlated with architectural design parameters.] (c) Relationship between type I-to-type II fibre ratio (left axis) and type IIA-to-type IIB fibre ratio (right axis).

With the exception of hybrid IIB fibres, muscle fibre types displayed a significantly lower CSA in the DS muscle than in both TM and DA muscles (Table 3). No significant difference (> 0.05) in size of fibre types was found between TM and DA muscles. As a consequence, the average CSA of all muscle fibre types was significantly higher in both TM and DA than in DS, but no significant difference was observed between TM and DA muscles (Table 3; Fig. 2b, right vertical axis).

The mean number of capillaries in contact with each fibre type was significantly lower for all fibre types, except types IIXB and IIB, in DS than in TM, and for types IIA and IIX in DS than in DA (Table 4). No significant differences were observed between TM and DA muscles, except that hybrid IIAX fibres had more capillaries in TM than in DA. The average number of capillaries per fibre was significantly higher in both TM and DA than in DS, but no significant differences were found between TM and DA muscles (Table 4).

The NADH-TR histochemical activity of all muscle fibre types was significantly lower in DS than in DA (Table 4). No significant difference (> 0.05) in this histochemical reaction was observed between TM and DA muscles. The NADH-TR activity of types I and IIA fibres was similar (> 0.05) in TM and DS, but this activity in both IIX and IIB fibre types was significantly higher in TM, almost doubling that observed in DS. Together, without considering fibre types, the average NADH-TR histochemical activity of all muscle fibres was significantly higher in DA and TM than in DS, but differences between DA and TM were not statistically significant (Table 4).

Discussion

This study focused on the structural organisation of llama muscles, with a dual purpose: first, to achieve a more thorough understanding of muscle design in this species; and second, to ascertain the potential correlation between muscle design and the characteristic locomotor behaviour displayed by this South American camelid as part of its process of adaptation to its habitat. With this aim in mind, the muscle architecture and fibre-type characteristics of three muscle sub-volumes that are primarily flexors of the shoulder joint (TM, DS and DA) were examined. Although potential regionalisation of muscle characteristics was not examined here (see below), an expected correlation between these two design parameters was observed between the two portions of the M. deltoideus (DS and DA), indicating a clear division of labour between these two sub-volumes. However, the most striking finding of this study was the unusual mismatch between muscle architecture and fibre types observed in the TM muscle. This muscle, whose architectural design was typical of a fast-contracting muscle (i.e. long fibres, low pinnation angles and a discrete PCSA) displayed fibre-type characteristics closer to those of a slow-contracting muscle (high I-to-II fibre ratio, large fibre sizes, large number of capillaries and high oxidative potential). Though unusual, this design of the TM muscle seems to be well suited to the characteristic locomotor behaviour displayed by the llama, which needs to increase ipsilateral stability during the support phase of the pace gait performed naturally by these animals (see below). Another major finding was that the skeletal muscle fibres displayed characteristics that presumably provide the llama with fatigue resistance and high force capability, these attributes being especially useful for an animal considered to be a good climber, used in the past as a beast of burden and well adapted for slow-to-moderate motion in high-altitude ecosystems.

The potential differences in muscle architecture and fibre types between regions within each muscle lay beyond the scope of the present study. Although there is now sufficient evidence that muscles are highly regionalised in many species (Kernell, 1998; Myatt et al. 2011), including llamas (Graziotti et al. 2004), and although the present results provide no indication of whether or not the muscles examined were homogeneous, there are several reasons for assuming that this constraint had a low impact on the results obtained. Firstly, the animals used were quite similar in body size and conformation; the data presented revealing very low variability (see SEM values in Tables 2–4). Secondly, extreme care was taken to standardise sampling sites within muscles from one animal to another. Nevertheless, further functional specialisations within the muscles examined here cannot be ruled out. The present results are limited, in that they address fibre-type comparisons among the three muscles at a very specific point of each muscle, and there may be significant variations in other portions of the muscles not examined in the present study.

The two portions of the M. deltoideus of the llama (DS and DA) showed clear differences in both muscle architectural design (Table 2) and fibre-type characteristics (Tables 3 and 4). The DS had longer fibres, with lower pinnation angles, as well as a higher percentage of fast-twitch fibres (including IIB), smaller fibres, and fibres with fewer capillaries and lower oxidative capacity than the DA. This evident correlation between architecture and fibre types, from which there are reported precedents (e.g. Hermanson, 1997; Gellman et al. 2002; Graziotti et al. 2004), suggests a division of labour between these two sub-volumes of the M. deltoideus (Higham & Biewener, 2011). The design of the DA muscle clearly implies a relatively high proportion of half sarcomeres arranged in parallel, which would optimise force production and fatigue-resistant properties (Roy & Ishihara, 1997). These features suggest a reduced or limited potential for excursion changes, but a significant role in maintaining joint position (stabilisation of the shoulder joint) against the force of gravity. By contrast, the design of the DS muscle implies a high proportion of half sarcomeres arranged in series (Roy & Ishihara, 1997), which might optimise displacement and high shortening velocity, suggesting significant excursions of this sub-volume in relation to dynamic flexion of the shoulder joint during the swing phase of the gait (Bodine et al. 1982; Hermanson, 1997).

To the best of our knowledge, there are no published studies specifically comparing fibre-type characteristics between the two portions of the M. deltoideus in quadrupeds. The architectural data obtained here in llamas largely agree with those previously reported for the same muscle portions in the European hare (Williams et al. 2007) and the racing greyhound (Williams et al. 2008). However, llama DA muscle fibres were considerably shorter (about 72% of the DS muscle fibre length) than in both the hare (about 47% of the DS muscle fascicle length) and the greyhound (similar length in both heads of the M. deltoideus). As the mass of the DA muscle is comparatively smaller in the llama (61% of the DS muscle mass) than in both the hare (52%) and the greyhound (43%), the resulting larger PCSA of the llama DA muscle emphasizes a clearly more postural (anti-gravitational) role and a greater ability to produce higher forces in this species than in other cursorial quadrupeds well adapted for high-speed gaits and non-steady-state locomotion. The EMG activity of the clavicular deltoideus muscle in the rat (functionally equivalent to the DA muscle of the present study) during the swing phase of locomotion is shorter and lower than that recorded for the DS muscle, indicating that its participation in dynamic movements of the shoulder joint is lower than that of the DS muscle (Hyland & Jordan, 1997). Furthermore, the differential function of the DA muscle of the llama compared with faster quadrupeds seems to be well matched with the natural ability of this camelid to pace at slow or moderate speeds (Van der Sluijs et al. 2010; Pfau et al. 2011), making the shoulder joint more stable when the whole-body weight is supported only by the ipsilateral limbs (see below).

The results of the present study provide limited information as to whether the DS muscle of the llama may perform an additional functional role besides its primary action as a flexor of the shoulder. Other EMG studies have shown that the DS muscle in rats (Hyland & Jordan, 1997) and humans (van der Helm, 1994) may be involved in both abduction of the arm and extension of the shoulder. Because the conformational and anatomical features of the llama (i.e. its narrow and evenly-contoured trunk) enable the limbs to swing freely forwards and backwards (Pfau et al. 2011), it may be postulated that some cranial regions of the DS muscle of the llama could be also involved in extension (i.e. protraction of the humerus) of the shoulder joint. Nevertheless, further research is required to confirm or discard this possibility.

When comparing TM and DS muscles, one might reasonably expect similar fibre-type characteristics in these two sub-volumes, given the similarity of their PCSAs (Table 2) and primary joint actions (flexion of the shoulder joint). However, consistent and significant differences in muscle fibre-type characteristics were observed between these two sub-volumes (Tables 3 and 4; Figs 2 and 3). The TM muscle fibre phenotype was clearly slower and more oxidative (more type I and fewer type IIB fibres, larger fibres with a higher number of capillaries and higher oxidative capacity on average) than the agonist DS muscle. Similarly, examination of fibre-type characteristics of the two muscles at the architectural extremes, TM (long fibres, low pinnation angle and small PCSA) and DA (short fibres, high pinnation angle and large PCSA), indicated that these two muscles exhibited no dramatic differences in terms of their constituent muscle fibre types (Tables 3 and 4; Fig. 2). There was only a relatively-modest significant difference in the proportion of type IIB fibres and in the capillarisation of type IIAX fibres. Although these muscles displayed similar fibre-type distribution at the specific sampling sites examined here, their fibre-type compositions may vary beyond these sample sites. Nevertheless, because the sampling point used for fibre typing in the TM muscle was located in the proximal sub-volume (Fig. 1A) and the architectural design of this sub-volume seems to be faster (i.e. longer fibres and lower pinnation angles) than that of the distal compartment (Fig. 2a), the likelihood of finding a faster fibre-type distribution in the distal compartment than in the proximal one is rather remote. Consequently, the discrepancies observed in the present study between architectural design and fibre types in llama TM muscle cannot be ascribed primarily to sampling site limitations. These discrepancies, already reported in many muscles in mice (Burkholder et al. 1994) and rats (Eng et al. 2008), clearly suggest that a muscle that is architecturally fast need not be physiologically and biochemically fast, and may reflect an obvious functional specialisation of single muscles (Eng et al. 2008). A review of several studies indicates that muscle architecture is certainly more complex than can be described using gross dissection alone. For example, the horse M. gluteus medius exhibits an evident fibre-type regionalisation as a function of sampling depth (Lopez-Rivero et al. 1992). However, though observing a wide range of fascicle lengths in this muscle (135–300 mm), Payne et al. (2005) did not observe a clear superficial to deep gradation of fibre length.

Figure 3.

 Serial cross-sections of the llama TM (left column), DS (middle column) and DA (right column) stained by enzyme histochemistry of myofibrillar ATPase (mATPase) after preincubation at pH 4.5 (a) and nicotinamide adenine dehydrogenase-tetrazolium reductase (b, NADH-TR), and by immunohistochemistry with the monoclonal antibodies A4-74-specific against myosin heavy chain (MyHC) isoforms IIA + IIX (c), S5-8H2 specific for MyHCs I + IIX + IIB (d), and BF-35 specific for MyHCs I + IIA. The fibres labelled I, A, X and B are ‘pure’ fibres containing MyHC I, MyHC IIA, MyHC IIX and MyHC IIB, respectively. Fibres labelled AX and XB are ‘hybrid’ fibres containing MyHC IIA plus MyHC IIX, and MyHC IIX plus MyHC IIB, respectively. See Table 1 for complete enzyme histochemical and immunohistochemical characterisation of these muscle fibre types.

The design of the llama TM muscle observed here suggests a dual function, involving both dynamic flexion of the shoulder joint during the swing phase of the gait and stabilisation of the limb during the stance phase, pulling the forelimb to the trunk. The long-fibred structure of this muscle means that it is well suited to fast contraction speeds and to producing forces over a greater range of motion, and is thus likely to play a dynamic role when the limbs swing forwards and backwards. However, this architectural design is not accomplished with a fast-contracting, glycolytic muscle fibre phenotype. This apparent incongruity can be partially explained, however, by the preferred footfall pattern of locomotion used naturally by these camelids. The slow-oxidative profile and large CSAs of fibre types observed in the TM muscle may reflect functional specialisation in order to increase the ipsilateral stability of the limbs during the support phase of the pace. Several EMG studies in other quadrupeds have demonstrated that this muscle is active during the support phase of the gait (Tokuriki, 1973; English, 1978), and that in addition to its primary action as a flexor of the shoulder joint, it can be considered as a complementary compartment of the M. latissimus dorsi that can also contribute additionally to adduction of the limb (Herring et al. 1993; Hyland & Jordan, 1997; Edge-Hughes, 2004).

Whereas muscle fibre size is directly related to its capacity to generate force (Powell et al. 1984), both capillary supply and the oxidative capacity of muscle cells endow the muscle with greater fatigue resistance during exercise (MacDougall et al. 1991; Green et al. 1998). In appendicular and trunk muscles, the average CSA of adult human muscle fibres ranges between 4000 and 5000 μm2, and between 2000 and 3000 μm2 in other species such as horses, ruminants, pigs and carnivores (see Table 5 for a comparative and overall estimation based on several earlier studies using similar methods). On average, the CSA of llama muscle fibres here was 5134 μm2, fibre sizes thus being similar to those of human fibres but clearly larger than those of other quadrupeds (Table 5). With the exception of the very resistant dog skeletal muscle (Acevedo & Rivero, 2006), the average capillary supply and oxidative capacity of llama muscle fibres were considerably higher than those of other species (Table 5). Moreover, whereas fibre-type composition in the llama shoulder flexors was quite similar to that previously reported in functionally equivalent muscles of both humans (Srinivasan et al. 2007) and primates (Singh et al. 2002), the slow-to-fast fibre ratio of these muscles was considerably higher in the llama than in other quadrupeds, including horses (Snow & Guy, 1980) and rodents (Alvarez et al. 2012).

Table 5.   Overall estimation of fibre CSA (μm2), number of capillaries in contact with each muscle fibre type (Cap/fib) and quantitative histochemical reaction of either succinate dehydrogenase or NADH-TR activity (OD, optical densities) in adult appendicular and trunk muscles of different mammalian species.
SpeciesCSA, μm2Cap/fibODReferences
  1. Values indicate means pooled for all fibre types from different muscles taking into consideration fibre-type composition.

  2. *Shoulder flexor muscles.

  3. Hindlimb muscles.

  4. CSA, cross-sectional area.

Llama*51347.150.39Present results
Llama33816.890.43Graziotti et al. (2001, 2004)
Human51904.580.29Green et al. 1998; Perez et al. (2002)
Horse31524.350.31Quiroz-Rothe & Rivero (2001)
Cow30553.530.36Moreno-Sanchez et al. (2008)
Pig28674.760.36Quiroz-Rothe & Rivero (2004)
Goat26623.100.35Arguello et al. (2001)
Thoroughbred22455.180.35Rivero et al. (2007)
Dog21166.370.55Acevedo & Rivero (2006)

Although the significance of these findings may not be fully explained in the present study, taken in conjunction they suggest that llama locomotor muscles may produce greater force and at the same time be more fatigue resistant than those of other mammal species. Furthermore, the fibre-type characteristics of the llama appear to be consistent with the mechanism through which this animal adapts to its particular habitat. The large-fibred, highly capillarised and slow-oxidative structure of the llama forelimb muscles appear to be well suited to an animal adapted for slow to moderate motion in a mountainous habitat at high altitude, which has long been used as a pack animal and is reportedly a good climber (Iñiguez et al. 1998; Van der Sluijs et al. 2010; Pfau et al. 2011). Recent specific research has indicated that adaptation to chronic altitude-induced hypoxia, combined with regular and prolonged physical activity, may increase skeletal myocontractile protein synthesis (Holm et al. 2010), rendering fibres larger than those observed at sea level, and leading to a fast-to-slow transition of muscle fibres, resulting in a faster activation of the mitochondrial oxidative metabolism (Doria et al. 2011). Moreover, although at high altitudes the acclimatisation-dependent increase in haemoglobin concentrations might well be sufficient to maintain adequate levels of muscular oxygen supply, and angiogenesis may not be necessary to preserve oxygen delivery to skeletal muscle (Lundby et al. 2004), human studies have reported increased capillarisation of muscle fibres as an adaptive mechanism to chronic hypoxic exposure (MacDougall et al. 1991).

Conclusion

In summary, this study highlighted an apparent mismatch between architectural design and fibre-type characteristics in three muscle sub-volumes that act synergistically in the flexion of the shoulder joint of the llama. Whereas the acromial and scapular parts of the M. deltoideus are architecturally and biochemically well designed for joint stabilisation during the support phase of the gait, and dynamic flexion of the shoulder during the swing phase of the gait, respectively, the TM muscle is architecturally designed for muscle excursion during the flexion joint, but biochemically is well adapted for the maintenance of static postural stability of the forelimb.

This apparent mismatch between the architectural design and fibre types of the llama TM muscle may reflect an obvious functional specialisation of this muscle in a species that, due to its preferred footfall pattern of locomotion (slow to moderate pace gait), needs to increase its ipsilateral stability during the support phase of the gait. Overall, the present results also demonstrate that the fibre-type characteristics of the llama limb musculature (high slow-to-fast fibre ratio, large fibre size, extensive capillarisation and high oxidative potential) enable greater force generation combined with improved fatigue resistance during exercise in comparison to other quadrupeds. These unique muscle features appear to be consistent with the characteristic locomotor behaviour of these New World camelids as part of their fascinating process of adaptation to a high-altitude habitat.

Acknowledgements

Contract grant number: Universidad de Buenos Aires, Ciencia y TécnicaV-039. Plan Andaluz de Investigación, Junta de Andalucía, Grupo CTS-179. The monoclonal antibodies A4-74, developed by Dr H.M. Blau, and BF-35, developed by Stefano Schiaffino, were obtained from the Developmental Studies Hybridoma Bank created under the auspices of the NICHD and maintained by The University of Iowa. The antibody S5-8H2 was a generous gift from Dr Eric Barrey, University of Evry, France. Two anonymous reviewers are thanked for their constructive criticisms.