Comparative forelimb myology and muscular architecture of a juvenile Malayan tapir (Tapirus indicus)

Abstract

The absence of preserved soft tissues in the fossil record is frequently a hindrance for palaeontologists wishing to investigate morphological shifts in key skeletal systems, such as the limbs. Understanding the soft tissue composition of modern species can aid in understanding changes in musculoskeletal features through evolution, including those pertaining to locomotion. Establishing anatomical differences in soft tissues utilising an extant phylogenetic bracket can, in turn, assist in interpreting morphological changes in hard tissues and modelling musculoskeletal movements during evolutionary transitions (e.g. digit reduction in perissodactyls). Perissodactyls (horses, rhinoceroses, tapirs and their relatives) are known to have originated with a four‐toed (tetradactyl) forelimb condition. Equids proceeded to reduce all but their central digit, resulting in monodactyly, whereas tapirs retained the ancestral tetradactyl state. The modern Malayan tapir (Tapirus indicus) has been shown to exhibit fully functional tetradactyly in its forelimb, more so than any other tapir, and represents an ideal case‐study for muscular arrangement and architectural comparison with the highly derived monodactyl Equus. Here, we present the first quantification of muscular architecture of a tetradactyl perissodactyl (T. indicus), and compare it to measurements from modern monodactyl caballine horse (Equus ferus caballus). Each muscle of the tapir forelimb was dissected out from a cadaver and measured for architectural properties: muscle‐tendon unit (MTU) length, MTU mass, muscle mass, pennation angle, and resting fibre length. Comparative parameters [physiological cross‐sectional area (PCSA), muscle volume, and % muscle mass] were then calculated from the raw measurements. In the shoulder region, the infraspinatus of T. indicus exhibits dual origination sites on either side of the deflected scapular spine. Within ungulates, this condition has only been previously reported in suids. Differences in relative contribution to limb muscle mass between T. indicus and Equus highlight forelimb muscles that affect mobility in the lateral and medial digits (e.g. extensor digitorum lateralis). These muscles were likely reduced in equids during their evolutionary transition from tetradactyl forest‐dwellers to monodactyl, open‐habitat specialists. Patterns of PCSA across the forelimb were similar between T. indicus and Equus, with the notable exceptions of the biceps brachii and flexor carpi ulnaris, which were much larger in Equus. The differences observed in PCSA between the tapir and horse forelimb muscles highlight muscles that are essential for maintaining stability in the monodactyl limb while moving at high speeds. This quantitative dataset of muscle architecture in a functionally tetradactyl perissodactyl is a pivotal first step towards reconstructing the locomotor capabilities of extinct, four‐toed ancestors of modern perissodactyls, and providing further insights into the equid locomotor transition.

In this study we reveal (1) the infraspinatus of Malayan tapir possesses dual originations, previously only reported in suids, which may be a feature exhibited by early tetradactyl equids (horses); (2) supraspinatus and extensor carpi lateralis contributions to limb muscle mass in the Malayan tapir are more than twice those for the same muscles in the monodactyl horse, reflecting shifts in muscle reliance during the evolutionary transition from tetradactyly to monodactyly in the equid forelimb.

Introduction

The Malayan tapir (Tapirus indicus Desmaret) represents the largest of the four widely accepted extant tapir species (Quse & Fernandes‐Santos, 2014). Malayan tapirs are considered to have diverged from the lineage which led to the modern neotropical tapirs approximately 25 Mya (Steiner & Ryder, 2011; MacLaren et al. 2018). The cranial and postcranial elements have for many years been known to differ from those of neotropical taxa (Earle, 1893), with recent quantitative analyses demonstrating clear divergences between Malayan and neotropical tapir osteology (MacLaren & Nauwelaerts, 2016, 2017; Dumbá et al. 2018). Morphological comparisons of the forelimb anatomy strongly suggest that the Malayan tapir possesses obligate function of its lateral fifth digit (Earle, 1893; MacLaren & Nauwelaerts, 2017; MacLaren et al. 2018), akin to some of the earliest extinct tetradactyl (four‐toed) perissodactyls, such as Propalaeotherium (Palaeotheriidae), Lophiodon (Lophiodontidae), Palaeosyops (Brontotheriidae) (Gregory, 1929; Holbrook, 2009; Franzen, 2010), and the ancestors of modern equids (e.g. Sifrhippus; Froehlich, 2002; Wood et al. 2011; Secord et al. 2012).

Unfortunately for the study of extinct vertebrates, muscular and ligamentous remains are almost invariably lost during the fossilisation process. As a result, the muscular arrangements and physiological cross‐sectional areas (PCSA), both critical inputs for modelling skeletal processes such as feeding and locomotion, can only be estimated based on available phylogenetic bracketing (Witmer, 1995). With regard to locomotion, the size of skeletal muscles and the arrangement of fascicles within them can be used to identify the potential velocity and power of muscle contractions (Olson et al. 2018); greater mass of proximal muscles relative to distal muscles can indicate reduced distal inertia during swing phase of running, affecting the kinematics of the animal (Wickler et al. 2004); and comparing PCSA between homologous muscles in different species can highlight differences in maximum force‐generating capacity (Payne et al. 2004). All these properties can aid in predicting maximum force and power muscles can generate, which are essential for musculoskeletal models examining changes in locomotor abilities (e.g. the equid transition; MacFadden, 2005). To further understand the transition from tetradactyl (four‐toed) forelimbs to the modern monodactyl (one‐toed) condition in equids, myological and skeletal information from their closest relatives will be of great value. Such functional myological data is readily available for derived equids (e.g. domestic horse Equus ferus caballus; Payne et al. 2004; Liebich et al. 2007; Watson & Wilson, 2007). However, for understanding extinct tetradactyl perissodactyl locomotion, modern tetradactyl perissodactyls (i.e. tapirs) may be considered more appropriate musculoskeletal analogues than modern, monodactyl horses. Within the four modern tapir species, the Malayan tapir (Tapirus indicus) represents the most functionally tetradactyl taxon (Earle, 1893; Bressou, 1961; Steiner & Ryder, 2011; MacLaren & Nauwelaerts, 2016, 2017). This study draws on the existing tapir literature describing muscular arrangements (Murie, 1871; Campbell, 1936; Bressou, 1961; Pereira, 2013) to inform a systematic dissection and muscular analysis of a juvenile Malayan tapir. In the absence of functional muscular data for tapirs, we quantified muscular masses and PCSAs for all the major forelimb muscles, compared them with a modern monodactyl relative (Equus), and evaluated the potential functional implications in future biomechanical modelling of ancestral hippomorphs (equids and relatives) and tapiromorphs (tapirs, rhinos and relatives). The forelimb was chosen for this comparison because it is tetradactyl in modern tapirs, as it was in the earliest equids. In addition, musculoskeletal modelling of equid limbs have focused on the forelimb (e.g. Swanstrom et al. 2005); therefore, forelimb data collected for tetradactyl perissodactyls will be of greater value for comparative modelling in future studies.

The specific action of a muscle during locomotion and gravitational support can be influenced by several characteristics, including activity pattern, fibre type and muscular architecture (Biewener & Roberts, 2000). The attributes of muscles that contribute to their mechanical action (i.e. their muscular architecture) can be described using several parameters; these include pennation angle (arrangement of fibres within the muscles relative to the axis of force generation) and PCSA (cross‐sectional area of the muscle with regards to its physiological properties). To our knowledge, these parameters are investigated for the first time in the forelimb of the genus Tapirus in this study. The upper forelimb muscles of tapirs are known to correspond to those of modern equids (Campbell, 1936; Bressou, 1961; Barone, 2000; Pereira, 2013). However, the highly specialised nature of the equid distal limb causes direct comparisons to distal tapir myology to be more problematic, although tendon and ligamentous attachments sites on the single functional and two ancillary digits of equids remain constant (Campbell, 1936; Diogo et al. 2015). We hypothesised that muscles in the upper forelimb of Tapirus would demonstrate similar patterns of muscular architecture (e.g. % mass, PCSA) to the upper forelimb of Equus. Conversely, we predicted large differences between the genera in muscle architecture in lower forelimb muscles, most notably for muscles that shift in function between tetradactyl (Tapirus) and monodactyl (Equus) species, such as the lateral digit flexors and extensors.

Methodology

Specimens

The two forelimbs of a juvenile Malayan tapir from the Koninklijke Maatschappij Dierkunde Antwerpen (KMDA) became available for study in 2016. The juvenile, a female approximately 5 months old, died from rapid onset viral encephalomyocarditis (Vercammen et al. 2017). The myocardial infection which caused the death of the juvenile tapir did not cause prolonged sickness or associated muscular atrophy. Medical conditions associated with captive perissodactyls which may affect skeletal muscle architecture and mass (e.g. obesity, capture myopathy, stress atrophy) have not been reported in tapirs (Duncan, 2018). Additional comparative material became available from an elderly male Malayan tapir, also from the KMDA; the left forelimb was examined (but not dissected) following its death due to natural causes. Dissection was not possible as this animal cadaver was being used for another, invasive experiment. An isolated manus of an adult lowland tapir (Tapirus terrestris) was also made available for visual inspection and comparison with the manus of T. indicus. We were unable to record muscular architecture information for either the adult male (limb used in different experiment) or the T. terrestris manus (muscle degradation), although visual examination of key muscle arrangements enabled qualitative interpretations. The dissection of the forelimbs took place over 4 weeks in the spring and autumn of 2016, performed at the Laboratory of Applied Veterinary Morphology of the Universiteit Antwerpen (Gebouw U, Campus Drie Eiken, Wilrijk, Antwerp).

Muscle architecture

Each forelimb muscle was isolated from the surrounding tissues and removed from the cadaver, following confirmation of origination and insertion sites. Where there were multiple heads to the muscle (e.g. triceps brachii), the locations of each head of the muscle were determined prior to removal, whereupon each head of the muscle was weighed and measured. The anatomy, attachment sites and functional action of the forelimb muscles under investigation can be found in Figs 1 and 2, and Table 1. Measurements were made for the following parameters: muscle belly length; tendon length; fascicle length; fascicle pennation angle. Tendon and muscle belly lengths were measured using a flexible ribbon tape, with incisions made into the muscular flesh to identify the extent of the tendons into the muscle. Muscle mass was measured using an OHAUS Scout‐Pro (SPU 602) measuring balance with 0.01 g precision; muscles weighing in excess of 200 g were weighed in several pieces and the combined total used. Tendinous tissue was carefully removed from the muscle following pure muscle mass measurement, with the mass of the tendinous tissue subtracted from the total muscle mass. MTU and muscle belly masses were both taken to determine which muscles exhibited the highest percentage of tendon mass. Muscle fascicle lengths were recorded along the line of fibre direction by measuring the distance from the fibre origin at the tendon to the end of the muscle fibre. Muscle and fascicle lengths were recorded using a ribbon tape, with repeated measurements (four to eight) being performed at regular intervals along the length of the muscle belly to generate a representative average length per muscle. Pennation angle to the tendon was calculated using a transparent protractor, again with repeated measures to provide an overall average pennation angle. Resting pennation angles of less than 5° were given a zero value (after the methods of Brown et al. 2003). Pennation angles were therefore not recorded for subclavius, teres major, medial triceps brachii, anconeal, pronator teres, brachioradialis, extensor digitorum communis (EDC) or extensor digitorum lateralis (EDL). All measurements were recorded on both right and left forelimbs, with overall averages representative of the individual.

Figure 1.

Muscles of Tapirus indicus right forelimb in lateral aspect. Photograph of muscles (A) with annotated diagram (B). Muscles: (a) m. cleidobrachialis; (b) m. subclavius; (c) m. supraspinatus; (d) m. infraspinatus; (e) m. deltoiudeus pars scapularis; (f) m. triceps brachii caput longum, (f′) caput laterale; (g) m. brachialis; (h) m. brachioradialis; (i) m. extensor carpi radialis (ECR); (j) m. extensor digitorum communis (EDC); (k) m. extensor digitorum lateralis (EDL); (l) m. extensor carpi ulnaris (ulnaris lateralis in equids) (UL); (m) m. flexor carpi ulnaris caput ulnare (ulnar head of FCU).

Figure 2.

Muscles of Tapirus indicus right forelimb in medial aspect. Photograph of muscles (A) with labelled diagram (B). Muscles: (a) m. subclavius; (b) m. subscapularis; (c) m. teres major; (d) m. brachialis; (e) m. coracobrachialis; (f) m. biceps brachii, (f′) lacerta fibrosus; (g) m. triceps brachii caput mediale, (g′) caput longum; (h) m. extensor carpi radialis (ECR), (h′) insertion tendon of ECR; (i) m. abductor pollicis longus (APL), (i′) insertion tendon of APL; (j) m. pronator teres; (k) m. flexor carpi radialis (FCR); (l) m. flexor digitorum profundus caput humerale (FDP); (m) m. flexor digitorum superficialis caput humerale (FDS), (m′) caput ulnare. FDS (l) and FDP (m, m′) closely combined after separate origination sites (see Table 3).

Table 1.

Origin, insertion and functional action of forelimb muscles of juvenile Tapirus indicus

Muscle	Abbreviation	Origin	Insertion	Function
Subclavius	SBC	Cranial angle of scapula (dorsal), extending to rib cartilage	Fascia of Pectoralis profundus	Supports and fixes the scapula; shoulder extension
Supraspinatus	SUP	Supraspinous fossa	Greater and lesser tubercules of humerus	Shoulder extension and support
Infraspinatus	INF	Main head: Infraspinous fossa Accessory head (see text): partial on the supraspinous fossa	Greater tubercle of humerus	Supports shoulder as lateral collateral ‘ligament’; shoulder flexion (after initial flexing)
Subscapularis	SUB	Dorsal border of scapula (medial)	Lesser tubercle of humerus	Supports shoulder; medial collateral ‘ligament’
Deltoideus	DEL	Tuber spinae	Deltoid tuberosity	Shoulder flexion
Teres major	TMJ	Caudal border of scapula (medial)	Teres tuberosity (deep to Latissumus dorsi)	Shoulder flexion
Teres minor	TMN	Caudal border and neck of scapula	Humeral shaft between greater tubercle and deltoid tuberosity	Shoulder flexion
Brachialis	BR	Distal to lesser tubercle of humerus	Cranial surface of radius	Elbow flexion
Coracobrachialis	CBR	Coracoid process (medial)	Distal humeral shaft (medial)	Upper arm adduction
Biceps brachii	BB	Supraglenoid and coracoid processes	Lacerta fibrosus (and coronoid process of ulna)	Elbow flexion; shoulder extension and stabilisation (during stance phase)
Triceps brachii caput longum caput laterale caput mediale	TBlo TBla TBme	Long: caudal angle and scapular border Lateral: ridge proximal to deltoid tuberosity Medial: medial shaft of humerus	Long: caudodistal part of olecranon Lateral: lateral olecranon Medial: medial epicondyle and anconeal process	Shoulder extension (caput longum) Elbow extension and stabilisation
Anconeus	ANC	Distal humeral shaft and olecranon fossa	Anconeal process and cranial olecranon	Elbow extension
Pronator Teres	PT	Proximal margin of medial epicondyle of humerus	Centre of radial shaft (medial)	Lower forelimb pronation
Brachioradialis	BRA	Supracondylar ridge (lateral) of the humerus	Styloid process of radius (medial)	Lower forelimb supination
E. d. communis	EDC	Supracondylar ridge and lateral epicondyle of humerus	Dorsal surface of all four ungual phalanges	Carpus and digital extension (all digits)
E. d. lateralis	EDL	Lateral epicondyle of humerus (distal to EDC)	Lateral digit IV and dorsal digit V	Digital extension (digit IV and V)
E. carpi ulnaris (Ulnaris lateralis)	UL	Lateral condyle of humerus (distal to EDL)	MC V (dorsolateral)	Metacarpal extension (MC V)
E. carpi radialis	ECR	Trochlear ridge, radial fossa and lateral epicondyle of humerus	Proximal MC III (dorsal)	Carpus extension and fixation
Ab. pollicis longus	APL	Interosseous space between radius and ulna	Lateral MC II and trapezium (second carpal)	Digit II abduction (modified from digit I abductor)
F. carpi ulnaris	FCU	Humeral head: medial epicondyle of humerus Ulnar head: medial olecranon	Posterior pisiform (accessory carpal)	Carpus flexion
F. carpi radialis	FCR	Medial epicondyle of humerus	Palmar surface of MC III	Carpus flexion
F. d. superficialis + F. d. profundus	FDS + P	Combined heads (Campbell, 1936) Ulna head: medial olecranon Humeral head: distal medial epicondyle Radial head: interosseous space between radius and ulna	FDS: intermediate phalanges II, III and IV FDP: ungual phalanges of digit II, II and IV; ulnar head shares fibres with UL insertion on digit V	FDS: forefoot and digital flexion; metacarpo‐phalangeal joint support FDP: Digital joint flexion
Ab. digiti minimi	ADM	Distal surface of ‘spatulate’ process of the pisiform	Lateral proximal surface of proximal phalange of digit V	Abduction of lateral fifth digit

E. = extensor; F. = flexor; Ab. = Abductor; d. = digitorum; apo. = aponeurosis; MC = metacarpal.

Muscular volume was calculated by dividing the mass of the muscle by mammalian muscle density constant (1.06 cm⁻³; Mendez & Keys 1960), following previous studies (e.g. Brown et al. 2003; Michilsens et al. 2009; Wareing et al. 2011; Böhmer et al. 2019). The estimated muscle volume was then divided by the average fascicle length for the muscle to generate the PCSA:

$$\text{PCSA}=\frac{\text{muscle volume}{\text{(cm}}^3)}{\text{muscle fascicle length (cm)}}$$

PCSAs were compared between the upper and lower limb muscles. Upper limb muscles were determined as those muscles intrinsic to the forelimb and which act upon the shoulder and elbow joint (Table 1). Lower forelimb muscles are designated as muscles which act upon the carpus and digits; several lower forelimb muscles cross the elbow joint, although their primary function is not the flexion or extension of the joint (e.g. brachialis, extensor digitorum communis). Data on Equus muscle architecture were taken from published sources (Brown et al. 2003; Payne et al. 2005; Watson & Wilson, 2007), and comparisons were therefore limited to the muscles which have architecture reported for equids. Muscle mass and PCSA data from those available muscles were compared with corresponding muscles of the dissected Tapirus. These muscles included the subclavius, supraspinatus, biceps brachii, triceps brachii (three heads), abductor pollicis longus (APL), combined flexor digitorum superficialis and f. d. profundus (FDS + P), extensor digitorum communis (EDC), EDL, flexor carpi ulnaris (FCU), flexor carpi radialis (FCR), extensor carpi radialis (ECR) and extensor carpi ulnaris (ECU). The ECU is homologous to the ulnaris lateralis (UL) in equids, which has lost its extensor function (Barone, 2000; Table 1). Published literature sources for comparative data of Equus included Brown et al. (2003) for the lower forelimb (APL, FDS + P, EDC, EDL, FCU, FCR, ECR and ECU), and Watson & Wilson (2007) for the upper forelimb (supraspinatus, biceps brachii, triceps brachii). PCSA from other ungulates was also used for comparisons across the entire forelimb (Rangifer, Wareing et al. 2011; Sus, Matthewson et al. 2011; Capra, Gewaily et al.

2017).

Statistical comparisons

As equids are much larger than juvenile tapirs, direct comparison of muscle mass was considered unsuitable. Ontogenetic studies of mammalian and avian muscle and tendon anatomy suggest that limb muscles can scale both allometrically and isometrically through ontogeny, dependent upon the muscle and the taxon (Miller et al. 2008; Olson et al. 2018; Martin et al. 2019). Unfortunately, no data is presently available for ontogenetic variation or scaling in muscle architecture for perissodactyls, and therefore scaling of juvenile Tapirus data to an adult size was not performed. We therefore compare patterns of mean % muscle mass and mean PCSA between Tapirus and Equus, rather than statistically comparing absolute or log‐transformed values.

Muscle mass and calculated PCSA for five upper forelimb muscles of Equus were taken from Watson & Wilson (2007) and for nine lower forelimb muscles from Brown et al. (2003). In the absence of total body mass data for all subjects in the analysis, individual muscle masses were calculated as a percentage of the sum of available limb muscle mass. The comparative dataset for Equus muscles included supraspinatus, biceps brachii, triceps brachii (three heads), flexor digitorum superficialis, flexor digitorum profundus (combined as the FDS + P), flexor carpi ulnaris (FCU), flexor carpi radialis (FCR), extensor digitorum communis (EDC), EDL, extensor carpi radialis (ECR) and extensor carpi ulnaris (UL; ulnaris lateralis in equids, where it has lost its extensor function; Brown et al. 2003). Percentage (%) mass for each tapir muscle was compared with that of Equus to inspect whether Tapirus % muscle mass fell within the standard deviations exhibited by the forelimb muscles of Equus.

Comparing pennation angles allowed for comparisons of potential force transmission to the tendons in the forelimb muscles (Brown et al. 2003). Within the tapir limb itself, one‐way analyses of variance (anovas) and Tukey‐B post‐hoc tests were performed to test for significant differences between pennation angles (following successful normality testing using Shapiro–Wilk test; Lang et al. 2012; Hady et al. 2015; Arruda et al. 2018). Significant differences in pennation angle were tested for between the lower forelimb flexors (FDS + P, FCU, FCR) and extensors (UL, ECR). Pennation angles of < 5° were excluded from anovas.

Results

Tapir muscle mass and volume

The triceps brachii (long head; 619.5 g), supraspinatus (444.9 g) and infraspinatus (315.5 g) represent the heaviest muscles in the upper forelimb of the tapir in this analysis. The combined deep and superficial digital flexors [flexor digitorum superficialis + profundus (FDS + P): 151 g], radial carpus extensor [extensor carpi radialis (ECR); 96 g] and common digital extensor [extensor digitorum communis (EDC); 53 g] were the heaviest muscles of the tapir lower forelimb (Table 2). Combined muscle volume of carpal flexors was substantially lower (47.2 cm³) than carpal extensors (128.9 cm³), whereas combined digital flexor volume (143.1 cm³) was double that of digital extensors (71.8 cm³). Upper forelimb muscles exhibited tendon masses between 0 and 20% of total MTU mass, whereas several lower forelimb muscles were recorded with tendon masses exceeding 33% of MTU mass (pronator teres, EDL, flexor digitorum superficialis + profundus; Table 2).

Table 2.

Muscular architecture of forelimb muscles of juvenile Tapirus incidus

Muscle	MTU mass (g)	Muscle mass (g)	Muscle length (mm)	Tendon length (mm)	Fascicle length (mm)	Pennation angle (°)
SBCa	208.6	207.7	34.5 ± 1	9.2	30.1 ± 1	< 10
SUP	464.3 ± 15	444.9 ± 1	27.4 ± 1	18.3 ± 1.8	9.3 ± 3	30 ± 1
INF	358.9 ± 1	315.5 ± 6	26.8 ± 2	33.3 ± 1	5.7 ± 1	37 ± 8
SUB	200.2 ± 24	183.8 ± 15	17.2 ± 1	14.4 ± 3	4.2 ± 1	38 ± 16
DEL	78.8 ± 1	77.6 ± 3	17.4 ± 3	16.3 ± 1	7.3 ± 3	16 ± 6
TMJ	143.5 ± 22	141.2 ± 19	22.7 ± 2	8.7	18.1 ± 1	< 10
TMN	19.8 ± 4	17.8 ± 2	10.3 ± 1	12.5 ± 1	5.3 ± 1	31 ± 5
BR	160.1 ± 9	157.5 ± 6	21.5 ± 1	12.5	13.7 ± 1	47 ± 7
CBR	14.4 ± 5	11.6 ± 3	14.9 ± 2	17.3 ± 1	2.2 ± 1	22 ± 7
BB	123.8	107.7	15.3 ± 1	26.2 ± 1	4.2 ± 1	35 ± 13
TBlo TBla TBmea	656.2 ± 22 203.8 ± 12 63.9	619.5 ± 15 197.8 ± 6 58.9	24.1 ± 2 21.7 ± 1 16.5	18.1 ± 1 14.0 12.2	9.9 ± 2 11.6 ± 2 10.9 ± 2	26 ± 9 29 ± 10 < 10
ANCa	53.4	48.4	13.8	7.2	9.1 ± 2	< 10
PTa	7.7	0.6	4.4	17.8	–	< 10
BRAa	25.0	22.9	22.0 ± 4.2	24.1 ± 3.7	20.9 ± 3.7	< 10
EDC	66.9 ± 11	53.7 ± 0.3	15.2 ± 1.6	35.5 ± 4.9	8.9 ± 2.2	< 10
EDLa	33.7	22.4	17.4	38.8	4.6 ± 0.5	< 10
UL	53.9 ± 12	40.5 ± 3.1	17.7 ± 0.4	27.5 ± 0.7	1.6 ± 0.5	41 ± 5
ECRa	110.3	96.2	20.0	33.5	9.5 ± 4.4	44 ± 3
APLa	25.6	21.1	11.0	22.5	2.7 ± 0.3	26 ± 8
FCU	22.2 ± 7	17.7 ± 4.8	18.6	23.0 ± 1.4	1.6 ± 0.6	37 ± 8
FCRa	37.2	32.3	16.3	28.2	3.2 ± 2.0	9 ± 5
FDS + FDPa	247.3	151.7	15.1 20.9	35.1 40.1	3.2 ± 2.8	11 ± 8

Measurements recorded on all muscles of left forelimb, with several muscle parameters recorded for right forelimb, with resultant measurements averaged (± measurement error). Fascicle length and pennation angles measured multiple times, with angles < 5° not recorded.

^aData recorded for only left forelimb.

Tapir muscle‐tendon unit (MTU) length

The longest MTU in the tapir upper forelimb was the infraspinatus (33.3 mm) and subclavius (34.5 mm), with the tendons of the biceps brachii, coracobrachialis, teres minor and infraspinatus exceeding muscle belly length (Table 2). The longest muscles in the lower forelimb include the brachioradialis, ECR and the FDS + P; the longest tendons belong to the digital extensors and flexors (Table 2).

Tapir pennation angle

The highest average pennation angles in the upper limb were recorded for the brachialis (47°), subscapularis (38°) and infraspinatus (37°), whereas the ECR (44°), extensor carpi ulnaris (= ulnaris lateralis; UL) (41°) and flexor carpi ulnaris (FCU; 37°) exhibited the steepest pennation angles for the lower forelimb muscles. The highest pennation angles in the entire forelimb were recorded for the fifth digit abductor (ADM; 55°) in the manus. anova between carpo‐digital flexor (FCU, FDS + P, FCR) and extensor (UL, ECR) pennation suggests that the extensor muscles of Tapirus are significantly more pennate than flexors (P < 0.01); anova results are presented in Table 3.

Table 3.

The anova results comparing mean pennation angles between autopodium flexor and extensor muscles in Tapirus indicus

Test	Sum Sq.	DF	Mean square	F	P	Perm. P
BG	919.811	1	919.811	13.93	0.003	0.004
WG	858.589	13	66.0453

BG = between group; WG = within group; Sum Sq. = sum of squares; df = degrees of freedom; perm. Perm P = permutational P‐value (10 000 replicates).

Comparative muscular parameters

Percentage muscle mass

Percentages of upper and lower forelimb muscle mass are compared between Tapirus and Equus in Fig. 3. Of the muscles available for direct comparison, the greatest percentage mass differences between Tapirus and Equus were observed in the supraspinatus, EDL and FCU (Table 4; Fig. 3). The long head of the triceps brachii accounted for the greatest percentage of muscle mass in the upper forelimb of both Tapirus and Equus, with the FDS + P and ECR constituting the largest percentages of muscle mass for the lower limb in both genera (Table 4; Fig. 3). The greatest percentage mass differences between Tapirus and Equus were observed in the supraspinatus, EDL and FCU (Table 4; Fig. 3).

Figure 3.

Comparison of individual muscle masses as a percentage of total forelimb muscle mass in Tapirus (white) and Equus (grey). Muscle abbreviations: SUP, supraspinatus; BB, biceps brachii; TBlo, triceps brachii (caput longum); TBla, triceps brachii (caput laterale); TBme, triceps brachii (caput mediale); APL, abductor pollicis longus; FDS + P, merged flexor digitorum superficialis + profundus; EDC, extensor digitorum communis; EDL, extensor digitorum lateralis; FCU, flexor carpi ulnaris; FCR, flexor carpi radialis; UL, extensor carpi radialis (in tapirs), ulnaris lateralis (in equids); ECR, extensor carpi radialis. Average % masses for upper limb muscles of Tapirus (this study) and Equus (from Watson & Wilson, 2007; n = 2); average % masses for lower limb muscles of Tapirus (this study) and Equus (from Brown et al. 2003; n = 7). Error bars denote standard deviation within Equus samples.

Table 4.

Comparison of percentage contributions of individual forelimb muscles to overall forelimb muscle mass between Equus and Tapirus

Muscle	Tapirus		Equus
	Muscle mass (g)	% Mass	Muscle mass (g)	% Mass
Total	2050.99	100	11175.74	100
SUP	444.95	23.81	1161.56	11.72
BB	107.74	5.76	663.66	6.69
TBlo	619.49	33.15	4059.17	40.95
TBla	197.79	10.58	887.53	8.95
TBme	58.88	3.15	159.94	1.61
APL	25.58	1.37	38.09	0.38
FDS + P	151.65	8.11	946.98	9.55
EDC	53.71	2.87	315.97	3.19
EDL	22.44	1.20	57.14	0.58
FCU	17.69	0.95	262.26	2.65
FCR	32.29	1.73	179.71	1.81
UL	40.51	2.17	364.07	3.67
ECR	96.18	5.15	814.75	8.22

Mass of individual forelimb muscles expressed as a percentage of total limb muscle mass for Tapirus (this study) and combined datasets from published literature for Equus (Brown et al. 2003; Payne et al. 2004; Watson & Wilson, 2007).

PCSA

Comparisons between PCSA in Tapirus and Equus forelimb muscles are presented in Fig. 4, with additional comparisons with selected artiodactyls in Table 5. PCSA in Tapirus is greatest for the long head of the triceps brachii (58.8 cm²), with the infraspinatus, supraspinatus, subscapularis and FDS + P all exhibiting relatively high PCSAs between 40 and 55 cm² (Table 5). The brachioradialis (supinator longus) exhibits the smallest PCSA for Tapirus in this study. When compared with other ungulate taxa, the juvenile tapir exhibits comparable PCSAs to adult Sus and Rangifer in rotator cuff muscles (infraspinatus, supraspinatus, subscapularis), although not in the teres minor, for which Tapirus exhibits a relatively low PCSA (3.2 cm²). The triceps brachii complex of Tapirus is very similar in PCSA to that of Rangifer (Table 5). When compared with Equus, the patterns of PCSA (rather than absolute values) for available muscles were in general similar (Fig. 4), with the greatest differences present in the biceps brachii and FCU (notably larger in Equus).

Figure 4.

Comparison of physiological cross‐sectional area (PCSA) in the forelimb muscles of Equus (squares; top) and Tapirus (diamonds; bottom). PCSAs for Equus taken from values in previous studies (Brown et al. 2003; Watson & Wilson, 2007); error bars denote standard deviation, with number of individuals presented above bar. Muscle abbreviations: SUP, supraspinatus; BB, biceps brachii; TBlo, triceps brachii (caput longum); TBla, triceps brachii (caput laterale); TBme, triceps brachii (caput mediale); APL, abductor pollicis longus; FDS + P, merged flexor digitorum sublimis + profundus; EDC, extensor digitorum communis; EDL, extensor digitorum lateralis; FCU, flexor carpi ulnaris; FCR, flexor carpi radialis; UL, extensor carpi ulnaris (in tapirs), ulnaris lateralis (in equids); ECR, extensor carpi radialis. PCSAs for upper limb muscles of Equus from Watson & Wilson (2007; n = 2) and average PCSAs from lower limb of Equus from Brown et al. (2003; n = 7).

Table 5.

Comparison of average physiological cross‐sectional area (PCSA) calculations for Tapirus against other extant ungulate taxa

Muscle	PCSA (cm2)
	Tapirus	n	Equus a	n	Rangifer b	n	Capra c	n	Sus d	n
SBC	6.5	1	23.0	7	14.3	1	–	–	–	–
SUP	45.3	1	150.3	2	46.5	1	4.9	3	31.2	4
INF	52.1	1	–	–	83.6	1	2.6	3	47.5	4
SUB	41.3	1	–	–	40.6	1	2.6	3	33.8	4
DEL	9.99	1	–	–	17.2	1	1.2	3	–	–
TMJ	7.4	1	–	–	5.9	1	7.0	3	–	–
TMN	3.2	1	–	–	11.1	1	8.4	3	8.7	4
BR	10.8	1	–	–	6.3	1	0.9	3	–	–
CBR	4.9	1	–	–	3.8	1	4.1	3	–	–
BB	24.1	1	244.8	2	46.1	1	1.0	3	–	–
TBlo TBla TBme	58.8 16.1 5.1	1	168.3 38.4 12.3	2	54.3 11.5 3.6	1	4.2 2.4 0.8	3	–	–
ANC	5.0	1	–	–	2.8	1	0.5	3	–	–
BRA	1.0	1	–	–	–	–	–	–	–	–
EDC	5.7	1	36.3	7	3.7	1	0.7	3	–	–
EDL	4.6	1	12.1	7	5.5	1	0.3	3	–	–
UL	24.7	1	193.8	7	73.9	1	0.8	3	–	–
ECR	9.6	1	99.3	7	27.7	1	1.1	3	–	–
APL	7.3	1	19.1	7	–	1	–	–	–	–
FCU	10.6	1	133.9	7	33.0	1	0.4	3	–	–
FCR	9.5	1	18.5	7	3.8	1	0.3	3	–	–
FDS + FDP	44.4	1	363.3	7	62.4	1	1.9	3	–	–

Dashes denote missing data (not recorded or absent in taxon). Specimens of Rangifer (n = 1) and Equus (n = 2) were sub‐adult.

^aBrown et al. (2003), Payne et al. (2004), Watson & Wilson (2007).

^bWareing et al. (2011) (sub‐adult).

^cGewaily et al. (2017) .

^dMathewson et al. (2014).

Discussion

The muscular arrangement, muscle mass and architecture of a Malayan tapir (Tapirus indicus) was here investigated and compared with similar architectural data available for the modern horse Equus ferus caballus. The arrangement of muscles in the forelimb of Tapirus is very similar to that of Equus, although the relative size (percentage mass) of certain muscles in the lower and upper forelimb differ between these genera. It would be remiss of the authors not to address the limitations of sample size, specimen age and comparative data availability. This study was conducted on juvenile material made available due to tragic and unfortunate circumstances, based on a non‐domesticated animal registered as endangered by the IUCN Red List (Momin Khan, 1997). Multiple specimens were therefore unavailable for dissection. Comparative material based on numerous equid individuals is well documented (Brown et al. 2003; Payne et al. 2004; Watson & Wilson, 2007), although a complete assessment of the muscular architecture of the Equus forelimb was unavailable. Despite the low sample sizes and specimen counts, observations made on the single tapir specimen offer a comprehensive account of forelimb muscular architecture in this genus, which to the authors’ knowledge has not been attempted until now. Within the tapir forelimb itself, several features were observed that were not previously noted in dissection reports and comparative studies (Murie, 1871; Campbell, 1936; Bressou, 1961; Pereira, 2013; MacLaren & Nauwelaerts, 2016, 2017). These are discussed below, along with comparative interpretations of both equid and tapir forelimb muscular architecture.

Forelimb musculature of Tapirus indicus

Within the forelimb of the tapir in this study, the largest muscles by mass and by PCSA include the triceps brachii, supraspinatus, infraspinatus and FDS + P (Tables 2 and 3). These muscles are heavily involved in propulsion and gravitational support, extending and supporting the shoulder (infraspinatus, supraspinatus), elbow (triceps brachii) and manus (FDS + P) (Liebich et al. 2007). When compared with other ungulate taxa (Equus, Sus, Capra and Rangifer; Table 5), the PCSA of the tapir deep lateral shoulder muscles (supraspinatus and infraspinatus) shows similarities to Rangifer and Sus, albeit with different absolute values. The ratio of the PCSAs of these muscles is more similar to Sus than to Rangifer, potentially explained by the relatively smaller attachment site (supraspinous fossa) in cervids and bovids compared with suids and tapirs.

The supraspinatus and infraspinatus have been shown to be of particular interest for tapir locomotion (MacLaren & Nauwelaerts, 2016), with the mountain tapir Tapirus pinchaque exhibiting a very large supraspinous fossa, interpreted as an adaptation for shock absorption at the shoulder (MacLaren & Nauwelaerts, 2016). In this study, the infraspinatus of T. indicus was observed not only to occupy the infraspinous fossa but also to pass over the scapular spine with an accessory head originating from the supraspinous fossa as well. This muscular arrangement is also observed in suids (Barone, 2000), with the attachment of the accessory head of the infraspinatus on the dorsal region of the scapular spine, in part explaining the peculiar morphology of the spine in T. indicus (as observed in MacLaren & Nauwelaerts, 2016). This muscular arrangement is observed in both juvenile and adult specimens of T. indicus, whereby the infraspinatus lies in the infraspinous fossa and a secondary head of the muscle originates from the dorsal scapular spine, distal to the ridge running along the scapular spine on the supraspinous side (MacLaren & Nauwelaerts, 2016). Previous studies did not report this morphology in T. indicus (Murie, 1871; Bressou, 1961); published muscular assessments of other tapirs are uncommon (Campbell, 1936; Pereira, 2013), and do not suggest this muscle arrangement is present in neotropical tapirs. None of the modern neotropical tapirs possess the clear ridge on the dorsal aspect of the scapular spine (MacLaren & Nauwelaerts, 2016). Assuming this scapular blade morphology is directly linked with the infraspinatus extending above the scapular spine in T. indicus (present in adult and juvenile specimens; MacLaren & Nauwelaerts, 2016), it seems likely that the infraspinatus of modern neotropical species is restricted to the infraspinous fossa. This is an avenue of investigation beyond the scope of this study but represents a clear and testable hypothesis for future comparative dissections and descriptions of muscle architecture in tapirs.

Unfortunately, no quantitative muscular architecture is available for other species of Tapirus, despite comparative dissections having been performed (Campbell, 1936; Pereira, 2013). When comparing muscular arrangements with those described for T. terrestris, we find differences to several origination and insertion sites reported in Pereira (2013); however, the majority of the muscle locations observed corroborate other myological investigations into both T. indicus and T. terrestris (Murie, 1871; Campbell, 1936; Bressou, 1961). Variation between these two taxa in the muscles of the manus may be expected due to the significant differences in carpal and metacarpal morphology (Earle, 1893; MacLaren & Nauwelaerts, 2017), possibly pertaining to different functional outcomes. From a brief comparison with a T. terrestris manus, we observed that the size of the digital interossei (flexores breves profundi manus; Campbell, 1936) in T. indicus were comparable to those of the adult T. terrestris (despite the young age of the T. indicus specimen). The most notable difference was found in the interossei of the fifth digit (IDV); this is the only interosseus of the manus that originates solely from the unciform. We observe that in T. indicus this muscle is broad, and (in our specimen) presented an additional belly to the muscle on the lateral side. Within the interossei of the T. indicus specimen, muscle pennation angles in the IDV were much lower than the other interossei (average IDII = 43°, IDIII = 30°, IDIV = 30°, IDV < 5°). The low pennation angle of the IDV suggests that the amount of force transmitted to the tendon of the IDV will be less affected by pennation angle than in the other interossei. By contrast, the fifth digit abductor (ADM) was the most pennate muscle in the forelimb (in this study). This indicates that the muscle force transmitted to the ADM tendon to abduct the fifth digit (i.e. spread the lateral digit) is high, though the velocity of movement is comparatively low due to the shorter fascicles arranged oblique to the line of muscle action. Further investigations into the comparative manus musculature between in the functionally tetradactyl T. indicus and other modern tapirs with less obligate use of the fifth digit (e.g. T. bairdii; MacLaren & Nauwelaerts, 2017) will doubtless provide further information on how, and potentially why, modern tapirs differ in their interaction with their underfoot substrate.

Forelimb muscle similarities between Tapirus and Equus

In this study, we put forward the hypothesis that patterns of muscular architecture in Tapirus and Equus would differ more in the lower forelimb than the upper forelimb, due to the highly specialised distal limb of Equus. We found this hypothesis to be only partially supported. Masses of lower forelimb muscles as a percentage of overall forelimb muscle mass in Equus differed from those of Tapirus in this study, constituting higher percentage contributions for half of the lower forelimb muscles (Table 4; Fig. 3). Across both upper and lower forelimbs, the greatest differences in % mass were represented by the abductor pollicis longus (APL; 256% larger in Tapirus), the flexor carpi ulnaris (FCU; 179% larger in Equus), EDL (108% larger in Tapirus) and the supraspinatus (SUP; 103% larger in Tapirus). Unfortunately, only five muscles were available for comparison between Tapirus and Equus in the upper forelimb, and these did not include the infraspinatus or subscapularis, which are intimately associated with the supraspinatus in the function of the shoulder (V; Mathewson et al. 2014). Nevertheless, with the muscles that were available for comparison, we show that 60% of upper forelimb muscle contributions to overall forelimb muscle mass exhibited by Tapirus fell outside the standard deviations of the Equus sample (Fig. 3; error bars). By contrast, 75% of lower forelimb muscle masses displayed by Tapirus fell outside the range of standard deviations from the % muscle mass calculations for the Equus sample (Fig. 3; error bars). This loosely supports our hypothesis of greater differences in patterns of muscle mass between Tapirus and Equus in the lower forelimb, although further examination of muscles across the forelimb of Equus will be necessary before any firm functional conclusions can be drawn.

PCSA results were not tested for significant differences, due to the large mass differences between these two taxa and comparison of a juvenile (Tapirus) with adults (Equus). Behavioural surveys of Malayan tapirs suggest that juveniles mature rapidly between 4 and 8 months, being weaned as early as 6 months of age (Gilmore 2001). However, as there are no current data on how ontogeny may affect perissodactyl muscular architecture, the potential issues that scaling up juvenile muscle mass or architectural properties may cause (e.g. isometric vs. allometric scaling in different muscles; Miller et al. 2008) precluded us from pursuing this as an option. Encouragingly, the limited number of previous studies investigating ontogenetic variation in architecture of mammalian limb muscles have shown that, although absolute values do differ, the patterns of PCSA across forelimb muscles do not vary greatly between adults and juveniles (Olson et al. 2018). Therefore, for the comparisons presented in this study, we believe that the juvenile tapir offers a viable approximation of the pattern of muscular architecture that may be observed in an adult of the same species. Trends in PCSA across the forelimb suggest that both the upper and lower forelimb muscles follow a similar pattern in Equus and Tapirus, which does not directly support our hypothesis. Two muscles show a clear deviation from the trend in PCSA between Equus and Tapirus: the biceps brachii and the flexor carpi ulnaris (FCU; Fig. 4). The more representative % mass results for the EDL and FCU (based on mass calculations for the entire lower forelimb) and the muscle architecture of the biceps brachii and FCU will be further discussed from a comparative functional standpoint.

The biceps brachii is proximomedially positioned in the upper forelimb, antagonistic to the triceps brachii, and one of the principal flexors of the elbow. The origination tendon of the biceps brachii passes from the coracoid process of the scapula through the intertubercular groove, passing along the long axis of the humerus to insert on the proximal radius (Liebich et al. 2007; Watson & Wilson, 2007). The biceps of Equus stores elastic energy within its internal tendon during the stance phase of locomotion to initiate limb protraction during swing phase (Wilson et al. 2003), which is of great value for a large animal running at consistent high speeds (e.g. equids). The relatively large PCSA of the biceps brachii in Equus indicates that the Equus biceps has the capacity to generate higher forces during limb protraction compared with Tapirus. This notable difference in PCSA may be explained by the dual role the biceps brachii plays in equid locomotion (Watson & Wilson, 2007), and the reliance on energy retention in flexor tendons for swift, efficient movement over long distances in modern equids (Wilson et al. 2003). By comparison, the supraspinatus of Equus has a much lower PCSA (150.3 cm²) than the biceps brachii (244.8 cm²), whereas Tapirus demonstrates the opposite condition (supraspinatus = 45.3 cm²; biceps brachii = 24.1 cm²). We may conclude from this that, in relative terms, the biceps brachii is more important than the supraspinatus as a shoulder extensor in Equus (as shown by Watson & Wilson, 2007), whereas the supraspinatus is of greater importance in this role for Tapirus. It is possible that this difference between shoulder extensor PCSAs has an ecological signal; open‐habitat sprinting cheetahs (Acinonyx) exhibit a large biceps brachii PCSA compared with the predominantly rainforest‐based jaguar (Panthera onca) (Cuff et al. 2016). However, although both comparisons demonstrate similar patterns of PCSA variation between species, establishing a true ecological signal would require more rigorous testing across a greater sample of taxa.

In accordance with our hypothesis, we observe notable differences in the lower forelimb of Tapirus when compared with that of Equus. As previously noted by several authors (Murie, 1871; Campbell, 1936), the flexor digitorum superficialis and flexor digitorum profundus in our specimen share sufficient muscle fibres to be considered combined (FDS + P) rather than separate as in equids, although the separate origination heads are homologous to those in Equus. We note that the extensor carpi ulnaris (ulnaris lateralis in Equus and many other ungulates) performs an extensor function rather than acting as a modified flexor, as in Equus (Brown et al. 2003); this corroborates previous observations on both T. indicus and T. terrestris (Murie, 1871; Campbell, 1936; Pereira, 2013). We postulate that the mean muscle volumes of carpal and digital flexor and extensors differ between tapirs and equids in part due to this extensor function of the ulnaris lateralis (UL). The flexor : extensor volume ratio for muscles acting on the carpus in Tapirus (0.37 : 1.00) is lower than that exhibited by Equus (0.95 : 1.00), and the overall flexor : extensor muscle volume ratio (0.95 : 1.00) suggests that the lower forelimb flexors are relatively greater in importance for Equus (1.43 : 1.00; Brown et al. 2003). This volume difference is partially explained by the relatively greater mass (and PCSA) of the FCU in Equus compared to Tapirus (Table 4; Figs 3 and 4). In turn, the greater % mass and PCSA of the FCU in Equus is in keeping with the necessity for energy retention in the stance phase and explosive release as the hoof leaves the ground during locomotion in equids. The rapid flexion of the carpus (and indeed the entire forelimb facilitated by the biceps brachii) expedites the initiation of the swing phase, a vital adaptation for running at high speed over large distances (Wilson et al. 2003).

The muscle with the lowest % mass and smallest PCSA in the equid lower limb is the EDL. While also having the smallest PCSA in Tapirus, this muscle has more than double the % mass for Tapirus than is observed in Equus. The function of this muscle goes some way towards explaining the differences in % mass for the EDL, as it is an extensor of the two lateral digits in Tapirus (digit IV and V; Murie 1871; Campbell, 1936; Bressou, 1961; Pereira, 2013). Equids, having lost full function of their fifth digit over 40 million years ago (MacFadden, 2005), are therefore likely to have reduced this muscle relative to those which still act upon the functional third digit (e.g. EDC). The insertion tendon which passed to the functional fourth digit in basal equids is retained in modern equids, and now inserts on the lateral surface of the medial phalanx of digit III (Liebich et al. 2007). As a result, we may conclude that tapirs with reduced functionality of the fifth digit compared with T. indicus (e.g. T. bairdii; MacLaren & Nauwelaerts, 2016) may exhibit a reduction in % mass in the EDL. Furthermore, this result has implications for modelling changes in equid locomotion through time. Greater understanding and quantification of muscular architecture in modern tapir species will be of great importance for estimating changes of muscular morphology and action during digit reduction through equid evolution.

Conclusion

In this study, we present the first published muscular architecture of the forelimb of the Malayan tapir (Tapirus indicus). We successfully quantify the muscles of the forelimb in this functionally tetradactyl perissodactyl using architectural measures including muscle mass, volume, pennation angle and PCSA. To our knowledge, this is the first attempt to quantify these data in the genus Tapirus, which remains enigmatic and somewhat understudied in its functional anatomy. Our investigation demonstrated that the shoulder musculature of Tapirus indicus shares greater similarities with suids than with equids in muscular arrangement and architecture (e.g. accessory infraspinatus, as in Sus). By comparing the muscular architecture of Tapirus with that of previously published studies on forelimb muscles, we identify multiple similarities between tapirs and their monodactyl relative Equus. Significant differences in % muscle mass accounted for by the supraspinatus, flexor carpi ulnaris and lateral digital flexor muscles were observed, in addition to relative divergence in PCSA of the biceps brachii and flexor carpi ulnaris. These differences graphically demonstrate several adaptations that equids have undergone to their muscular architecture during the progression from tetradactyl forest‐dwellers to monodactyl, open‐habitat specialists. When investigating locomotion in the small, closed‐habitat ancestors and relatives of modern equids (e.g. Sifrhippus, Hallensia, Propalaeotherium), it will be essential to understand how different muscle groups in the limbs have adopted greater or more reduced importance through evolutionary transitions such as that of equids. Results from this study are an ideal first step towards developing viable locomotor models of early equid ancestors. Further studies, hopefully adding more comparative material of neotropical tapirs and comprehensive assessments from modern Equus muscular architecture, may also focus on correlating differences observed in muscular architecture patterns with changes in bone morphology across locomotor transitions in Perissodactyls.

Author contributions

This study was proposed by J.M. Both authors collected experimental data, J.M. analysed data and designed figures. The manuscript was written by both authors (J.M. 67%, B.K.M. 33%).