«Measurements of Trackways as a Method for Assessing Locomotion in Dairy Cows Evgenij Telezhenko Thesis (Licentiate) Sveriges Lantbruksuniversitet ...»
Hitchcock’s works, in which thousands of tracks were systematically evaluated and classified, are still classic references for ichnology, the science of fossil trackways and traces. Trackway analysis offers unique and important insights into the behaviour of dinosaurs (Kuban, 1989) revealing details about the lives of ancient animals that are not obvious from skeletal remains.
In 1897 Bradford published his work in which human gait was examined by observation of footprints on sand, dust, mud and snow. In spite of remarkable progress in gait analysis, data obtained from footprints are likely to continue to be used in human biomechanics as a simple and inexpensive method to quantify important aspects of gait (Wilkinson, Menz & Raspovic, 1995). Data obtained from direct measurements of footprints are often used in human biomechanics for clinical investigations (Stolze et al., 1998) and to validate new and more sophisticated gait analysis systems (Gaudet et al., 1990; McDonough et al., 2001).
Still the measurements of trackways remain a fairly unusual method of analysis of animal locomotion including locomotion of dairy cows. Benz (2002) used foot imprints to measure the step length of cows on different floorings. However, there has been no detailed methodological study on use of the trackway measurements for evaluation of locomotion in dairy cattle.
The aim of this study was to assess the possibility to describe the locomotion of dairy cows using measurements of their trackways. The focus was on the question of reliability of measured parameters with special emphasis on suggesting indicators for cow locomotion comfort. In addition, the aim was to evaluate, by using the measurements, differences in locomotion in animals with different conditions of the locomotor system on different floors.
Summary of materials and methods
Animals and housing The study described in Paper I was carried out on the research farm of the Swedish University of Agricultural Science in Alnarp, Sweden. The 32 studied cows were of the Swedish Holstein breed, mainly in first and second lactation, without signs of lameness. The average height at withers was 143 cm (range 133–154 cm) and average body length was 181 cm (range 161–192 cm). The cows were kept in cubicles with slatted concrete passageways. An alley (with a slatted concrete floor) from the cubicles to the milking parlour was used for the gait analysis. The dynamic coefficient of friction (COF) of the slats was 0.55.
The study described in Paper II was carried out on a commercial organic dairy farm in western Sweden. Eighteen Swedish Red and White and 18 Swedish Holstein cows were studied. All cows were kept under similar management conditions in cubicles with slatted concrete walkways.
The presence and severity of lameness (Paper II) was scored on a four-point scale, modified from Sprecher et al. (1997), according to which, animals with normal gait (n = 25) and mildly (n = 6) and moderately (n = 5) lame cows were identified. No severely lame cows were used in the study. A walkway from the milking parlour to the pasture was used. The walkway included both slatted concrete (dynamic COF = 0.31) and solid concrete (dynamic COF = 0.58). Twenty mm thick KEN® rubber mats (Gummiwerk Kraiburg Elastik, Germany) with rubber-studded underside profiles were used on the slatted and solid 10 m of the walkway. The dynamic COF of the rubber mats was 0.46. The rubber mats were introduced 1 month before the measurements started. At the exit to the pasture a sand track was prepared from moist sand compacted by a tractor.
The friction of the floors in both papers was measured with a portable friction tester (PFT). Because of different measurement principles, the friction values obtained with the PFT were higher than those obtained with the more commonly used pendulum swing device (skid resistance tester (SRT) or British pendulum tester). To make the results more comparable to those of other studies, PFT friction values were converted to SRT friction values by a formula developed by the Swedish National Road and Transport Research Institute and the SRT values were then transformed into a COF value (Astrom, 2000).
Gait assessment In both papers a trackway measurement system was used for the gait analysis. To obtain visible foot imprints (Fig. 1), lime powder (Ca(OH)2) was dispersed over the walkway and mixed with a thin layer of slurry (Fig. 2). The cows walked one by one over the prepared surface at their own chosen speed. In order to make the cows walk continuously, a person followed them slowly at a distance of several metres (Fig. 3). Only trackways from cows that walked evenly were used for the measurements. In the first paper the following elements of trackways were measured (see Fig. 2 in Paper I): stride length (the distance between two consecutive imprints of the same rear foot); step angle (the angle between the lines connecting the three consecutive imprints of the rear feet); step width (the distance between the right and left rear foot imprints, along the line perpendicular to the line of progression); step length (the distance between two consecutive imprints of the left and the right rear foot); step asymmetry (the absolute value of the difference between two consecutive steps); overlap (the lengthwise distance between the front foot imprint and the next imprint of the same side’s rear foot);
step abduction-adduction (the sideways distance between the front foot imprint and the next placement of the same side’s rear foot).
Measurements in Paper II included all parameters mentioned above with the exception of step width and step abduction-adduction. The measurements were taken from four consecutive strides using a ruler and an angle meter (Fig. 4). The measurements and their reference points are described in detail in Paper I. The walking speed of each cow was measured with a stopwatch.
Figure 2. Preparing the walkway with Figure 3.
Obtaining the trackway and lime powder and slurry measuring the speed Locomotion measurements in Paper I were repeated after three weeks in all cows.
Cow locomotion in Paper II was tested on five different surfaces in the same test run, viz. solid concrete without rubber mats, slatted concrete with rubber mats, slatted concrete without rubber mats, solid concrete with rubber mats and wet, compacted sand.
All measurements were made by the author of the thesis (E.T.).
Figure 4. Measuring the trackway
Statistical analysis Paper I The analysis was performed using the PROC MIXED procedure in SAS (SAS v. 8 for Windows; SAS Institute, Inc., 2002). The statistical model for trackway measurements included effect of cow, test day and interaction between cow and day of the test. All the terms in the model were considered random and independent.
The parameters for the variance components were estimated with the residual (restricted) maximum likelihood (REML) method. To assess the impact of each factor on the total variance, the percentage of each variance component was calculated. The proportion of the total variance that is due to variance among cows was used as a measure of reliability. We interpreted proportion of between-subject variance 20% to mean “very low reliability”; one of 20–45% to represent “low reliability”; one of 46–70% to show “moderate reliability”; and one of 70% to mean “high reliability”.
To measure the strength of the linear association between gait parameters, Pearson product-moment correlation coefficients were calculated for individual means. The same procedure was applied to assess correlations of locomotion parameters with body size. To estimate the linear association between gait parameters within individuals, analysis of covariance was used and partial correlation coefficients within subjects were calculated according to the method described by Bland & Altman (1995).
Paper II The speed and individual trackway means, obtained from measurements of four consecutive strides on each floor, were calculated with analysis of variance (ANOVA) with the between- and the within-subject design. Two different statistical models were used to describe the data. The first model included floor as within-subject effect and lameness degree as between-subject effect. In the second model only cows without signs of lameness were included in the analysis. In this model breed and parity were included as between-subject effects, and floor as the within-subject effect. The cow effect was considered random and other effects were considered fixed in both models. Multiple comparisons were performed using Tukey’s Honestly Significantly Different (HSD) test. We used JMP v. 5 for Windows (SAS Institute, Inc., 2002) for the statistical analyses.
A 5 % significance level was used throughout both papers.
Summary of results
Paper I The reliability was moderate for walking speed (66.67%), stride length (62.35%), step angle (57.23%), step width (51.8%) and overlap (65.38%). Step length and step abduction-adduction showed a fairly low reliability (37.76% and 25.44%, respectively).
Step asymmetry had almost negligible between-subject (18.09%) and betweentest variation (5.30%) but within-test variation was large (76.57%).
Increased stride was associated with an increased step angle, as well as with increased step length and increased overlap. Increased abduction was associated with a decreased step angle, increased step width and slightly increased step asymmetry.
Height at the withers was positively associated with stride length and step angle and negatively with step width and abduction-adduction. Body length correlated positively with stride and step length and tended to have negative correlation with overlap.
Paper II Strides and steps were considerably shortened and the rear foot imprints were placed at a greater distance behind the front ones (larger negative overlap) on a slippery slatted concrete floor in comparison with all the other floors. On the solid, crude concrete floor the cows took shorter strides and steps than on the sand surface, but the speed did not differ significantly. Length of strides and steps increased on the elastic rubber mats compared with the corresponding concrete floors. In moderately lame animals speed was lower and the stride and step lengths were shorter compared with those in non-lame and mildly lame cows. Analyses within each lameness degree revealed that the moderately lame cows walked with a significantly wider posture on solid and slatted concrete than on the yielding surfaces. Comparisons of cows with different lameness scores within a floor type showed that the moderately lame cows compared with the non-lame ones had a smaller step angle on the concrete flooring while there were no significant differences between lame and non-lame animals when walking on yielding surfaces. Step asymmetry increased progressively from the non-lame to the moderately lame cows on all floor types. In cows with moderate lameness the step asymmetry was less pronounced on the sand and on the floor with continuous rubber covering.
Trackway measurements and error due to measurements Measurement of trackways is a method for obtaining several spatial kinematic variables in cows in their actual environment. The cow trackway measurements system was developed to analyse cows’ gait on different floorings (Telezhenko, Bergsten & Manske, 2002). Clearly the information obtained from trackways is not as precise and complete as that obtained with three-dimensional (3-D) motion analysis. However, the method of trackway measurements is inexpensive, easy to apply and interferes minimally with the movement of cows in their natural environment. Because it uses direct measurements, trackway analysis could result in more reliable records than achieved with e.g. 2-D video analysis because of deviation in animal movement from line perpendicular to the camera. However, the measurements of trackways lack some important kinematic parameters such as different angulations and trajectories of joints as well as important temporal stride characteristics. The only temporal parameter used in the present studies was speed, calculated from the time used for passing the tested walkways. By measuring the time with a stopwatch, certain measurement errors are inherited. To improve the precision of speed measurements, it would be possible to complete measurements of the trackways with a video recording of the cow passing the walkway.
Because all measurements were performed by one operator we did not assess between-operator variation. Nevertheless it should be noted that the source of such variation could lie in different determination of reference points for the measurements. Detailed recommendations for finding points of reference were made in Paper I. The main argument was that each place of the imprint could be a reference point; however, the reference point should consistently be the same place in each footprint within a trackway. It has been shown in humans that with very well-defined reference points for measurements of gait parameters from footprints, between-observer variability is very low (Wilkinson & Menz, 1997).
The abduction-adduction of the limb implies that the limb moves from or towards the mid-sagittal plane. In measuring the lateral deviation from the front imprints we did not take into account the abduction-adduction of the front limbs.
We made the assumption that the front limbs move more parallel to the midsagittal plane than the rear limbs (Ndikuwera & Zishiri, 1990). Because of low reliability of the trait abduction-adduction was not used in Paper II and its meaning was substituted with the parameter characterising base of support, viz.
Although it is possible to identify differences in locomotion on floors of very different quality, trackway measurements are probably a fairly rough method for determining smaller differences in locomotion on more similar surfaces.
Reproducibility of gait The reliability of gait parameters obtained from trackways and defined as portion of between-subject variation was lower than that of published values in trotting horses (Drevemo et al., 1980a; b, c) and humans (Kadaba et al., 1989). However, even variable gait patterns, measured under natural conditions, could be worth studying, since “steady locomotion is more the exception than the rule for animals travelling through natural environments” (Dickinson et al., 2000).