Basic limb kinematics of small therian mammals

The evolution of mammalian limbs is marked by the transition from a two-segmented, sprawled tetrapod limb to a three-segmented limb (for a review,see Fischer, 1999) and from lateromedial, undulatory movements to dorsoventral movements of the body axis(Hildebrand, 1974). The addition of locomotory active segments is achieved in different ways for fore-and hindlimbs: the shoulder blade becomes moveable and is added as the proximal segment to the `old' ancestral forelimb(Jenkins and Weijs, 1979),whereas on the hindlimb the existing distal element, the foot, is prolonged and becomes the third segment by the `new' ankle joint. As a consequence serially homologous elements such as humerus and femur no longer functionally correspond to each other. The functional correspondence is now: shoulder blade to thigh, upper arm to lower limb, forearm to foot. These postural changes influence fundamentally the action of the limbs during locomotion.

Our first aim is to present quantitative kinematic data of fore- and hindlimbs' movements for several, only distantly related, small therian mammals at different gaits. Based on this substantial amount of highly detailed work, a comparative study of limb configuration and kinematics was undertaken to look for basic kinematic similarities emerging together with the new therian limb. The principles that emerge from these studies, together with the published work especially on cats will be also tested for their validity in midsize ungulates.

Another key innovation of therian locomotion is the regular use of in-phase gaits (gallop, half-bound, bound); crocodiles show gallop only exceptionally as juveniles (Zug, 1974). According to Hildebrand(1985), sagittal spine movements occur typically in fast carnivores, lagomorphs, and rodents. We present first data for other especially small therians. The consequences of these `new' gaits on fore- and hindlimbs and especially lower spine kinematics have never been quantified using cineradiography. Based on the X-ray study of slow walking (i.e. exploratory walking) in the tree shrew Tupaia glis, a restricted bending region of flexion between Th11 and L1 has been described (Jenkins, 1974a). This observation will be tested in faster `in-phase' gaits. In our study, we also included animals with and without tails, to test their influence on the kinematics of the sagittal back movements.

Cineradiography is the only tool that recognizes the exact kinematics of all proximal skeletal parts hidden under the skin and subcutaneous fat. Previous studies on quadrupedal therian mammals quantitatively analysed either(1) single joints such as the shoulder joint in rats (Rattus norvegicus; Jenkins,1974b), hip joint and back movements in the skunk (Mephitis mephitis; Van de Graaff et al.,1982), trunk movements in the shrew-like opossum (Monodelphis domestica; Pridmore,1992), elbow and wrist joint in the potto (Perodicticus potto; Jouffroy et al.,1983), ankle joint in kangaroo rats (Dipodomys spectabilis; Biewener and Blickhan,1988), single limbs at specific gaits (e.g. hindlimb in cats Felis catus f. domestica; Kuhtz-Buschbeck, 1994), or (2)qualitatively single steps only at one or more gaits (e.g. Tupaia; Jenkins, 1974a; jirds Meriones shawi; Gasc,1993). Most data are available for the cat, which has long been used as a model organism (Engberg and Lundberg, 1969; Goslow et al.,1973; Miller and Van der Meché, 1975; Jenkins and Camazine, 1977; Sontag et al., 1978; English, 1978a, b, 1980; Halbertsma, 1983; Hoy and Zernicke, 1985; Caliebe et al., 1991; Kuhtz-Buschbeck et al., 1994;Boczek-Funcke et al., 1996, 1998, 1999). From outside of our group the only cineradiographic study on fore- and hindlimb kinematics at different gaits was published by Rocha Barbosa et al.(1996) on the domestic guinea pig Cavia porcellus.

Scapular displacement during quadrupedal locomotion has been measured in Felis (Miller and Van der Meché, 1975; English,1978a; Sontag et al.,1978; Boczek-Funcke et al.,1996), Rattus(Jenkins, 1974b), Cavia (Rocha Barbosa et al.,1996), Virginia opossum, Didelphis marsupialis(Jenkins and Weijs, 1979) and vervet monkeys Cercopithecus aethiops(Roberts, 1974; Whitehead and Larson, 1994). The studies named describe a clear pro- and retraction of the shoulder blade during locomotion, but the impact of this displacement on step length and the overall kinematics of the forelimb has never been determined.

We have collected kinematic data on eight different small therian mammals using cineradiography. Data of two phylogenetically distant metatherians(Dasyuroides byrnei, Monodelphis domestica) and six eutherians belonging to five different orders (Primates: Microcebus murinus;Rodentia: Galea musteloides, Rattus norvegicus; Lagomorpha: Ochotona rufescens; Hyracoidea: Procavia capensis;Scandentia: Tupaia glis) are now available to elaborate upon the kinematic principles of small mammal locomotion. In addition, data based on analyses of two artiodactyls, Tragulus javanicus and the domestic goat Capra hircus (Lilje and Fischer, 2001), a very small rodent (Acomys cahirinus),as well as another primate (Saguinus oedipus; Schmidt and Voges, 2001) are included in this paper (e.g. in illustrating touch-down and lift-off positions). Single kinematic studies including detailed information about metric parameters, footfall patterns and gait-specific kinematics, as well as intralimb timing are already published on Procavia capensis (Fischer, 1994, 1998), Ochotona rufescens (Fischer and Lehmann,1998), Tupaia glis(Schilling and Fischer, 1999),and Eulemur fulvus (Schmidt and Fischer, 2000). These published data are drawn together here by further calculations, for example, on the contribution of limb segments to step length.

Limb kinematics were studied by cineradiography in adult individuals of eight small therian species which belong to different higher-order taxonomic groups of mammals. The kinematic analyses are based on more than 80,000 digitised X-ray frames. Table 1gives an overview of the species under investigation by denoting number of individuals, body mass and body length. For sake of simplicity, we named the species by their generic names only. All experiments were registered by the Committee for Animal Protection of the State of Thuringia, Germany.

Individuals were positively conditioned to move on a horizontal motor-driven treadmill within a Plexiglas enclosure (length 100 cm, width and height were adapted to the requirements of each species) except for the arboreal quadrupedal Microcebus, which walked on a rope-mill, an arboreal analogue of a treadmill. Treadmill speed was not fixed, but the operator attempted to keep the running animal in front of the X-ray screen for as long as possible. Thus, the operator adjusted the speed to obtain certain preferred speeds of the animals. Comparisons of treadmill locomotion and unrestrained locomotion have shown that the basic schemes of kinematics are the same in both situations (Fischer,1999).

The X-ray equipment consisted of an automatic Philips® unit (Type 9807 501800 01) with one X-ray source image-amplifier chain. Pulsed X-ray shots were applied (approximately 50 kV, 200 mA). The X-ray images on the image intensifier were recorded either on 35 mm film using an Arritechno R35-150 camera or with a high-speed CCD camera (Mikromak Camsys®) operating at 150 frames s^-1. The animals were filmed in a lateral projection with a maximum exposure time of 10 s. As some of the animals were larger than the area of interest covered by the image-amplifier (20.5 cm×15.0 cm), fore-and hindlimbs were recorded separately. An orthogonal wire grid, perpendicular to the projection plane, provided reference points for correction of geometrical distortions and metrical calculations.

X-ray films were copied onto video tapes and A/D-converted using a video processing board (Screen Machine® I, Fast® Multimedia AG, Munich,Germany), and further analysed by application of the software `Unimark 3.6'(by R. Voss). This software makes it possible to digitise interactively previously defined landmarks with a cursor function; it also corrects distortions automatically and calculate angles and distances. The positions of digitised landmarks and angles calculated in the parasagittal plane are illustrated in Fig. 1. Angles calculated are the projections of angles onto the sagittal plane representing their contribution to movements in the plane of forward motion.

The errors generated by digitisation of skeletal landmarks and their influence on the angles calculated were tested by repetitive digitisation(five times) of one sequence (25 frames) for each species. The digitisation error depends on the size of the animal and the image contrast of skeletal elements. It ranges from 0.5° to 2.0° for segment angles (see below)and is roughly 1.0-3.0° for joint angles, because the errors of two adjacent segment angles combine in joints following the Gaussian rules of error propagation.

Limb joint angles were defined anatomically and measured at the flexor side of each joint. Segment angles were calculated versus the horizontal plane. We shall use the term protraction (= cranial rotation) for the cranial displacement of the distal end of each segment. Retraction (= caudal rotation)describes its caudal displacement. Maximum amplitudes of joint excursion during stance and swing phases were calculated from the initial moments of segment and limb-joint movements. Effective angular displacements (EAD) were defined as differences of angles at touch-down and lift-off. The ratio EAD versus maximum joint amplitudes gives the coefficient of stance phase(CSP). A ratio higher than 0.5 indicates a joint's action resulting in a forward propulsive movement.

Fischer and Lehmann (1998)proposed an `overlay method' to calculate the relative contribution of angular movements to step length. While the CSP indicates the non-propulsive vertical work of joints, only the overlay method enables calculation of the relative contribution of segment movements to horizontal forward motion, because it considers the displacements of pivots of the limb segments during stance phase.

Summarising the `overlay method' in short, the calculations are based on mean values of typical gait sequences, of which stance and swing phase duration are scaled to equivalent relative durations using the method of linear interpolation. A polynomial fit of sixth order is used to interpolate data. For calculation, angular values are defined in the vertical plane to be positive if the distal end of the segment is in front of the proximal end. The horizontal distance (l_p) between tip of toe and the pivot of the whole limb is determined for every single limb configuration during stance phase, using the lengths of segments and their angular excursions against the vertical plane. By overlaying the proximal segment onto the next configuration, without changing angles in the more distal joints, the difference between the horizontal excursion at instant i(l_pi) and at instant i+1 (l_pi+1) is the step length caused by the rotation of each particular segment. For each segment the absolute contribution to step length is given by the summation of all single-frame calculations in stance phase. Finally, the contribution of the remaining segments to forward motion is calculated in the same way, except for the subtraction of the angular movement achieved by the sagittal rotation of the more proximal segment(s). The relative contribution of segment displacement to step length depends on the pivot's height, the effective angular displacement (EAD) and the length of the segment.

Fore- and hindlimb kinematics were studied by cineradiography in eight small therian species. At least 15, but up to 300, step cycles (period between two successive touch-downs of one limb) were analysed within broad ranges of running speeds (Galea, 0.3-1.3 m s^-1, Dasyuroides, 0.3-1.0 m s^-1, Microcebus, 0.3-1.7 m s^-1, Monodelphis, 0.4-2.4 m s^-1, Procavia, 0.2-2.1 m s^-1, Ochotona, 0.2-1.6 m s^-1, Rattus, 0.4-0.8 m s^-1, Tupaia,0.5-1.6 m s^-1). Each species preferred different gaits within even the same speed ranges; e.g. Rattus and Microcebus only used`walk' but pikas only used `halfbound' and `gallop'. Because most animals changed both frequently and suddenly between walk and trot, or gallop and halfbound, we pooled walk and trot data as symmetrical gaits, as well as gallop and halfbound as in-phase gaits.

Kinematic data comprise segment and joint angles at touch-down and lift-off, maximum amplitude of the stance phase, contribution of segment displacement to step length and the coefficient of stance for limb joints at symmetrical gaits (Tables 2A, 3A) and in-phase gaits (Tables 2B, 3B,C). Differences between trailing and leading limbs (first and second touch-down) were observed on the hindlimbs only at gallop and half bound and therefore, these limbs are presented separately.