Introduction
Swimming is a practical, safe exercise for dogs that minimizes joint pain and facilitates weight loss and recovery from orthopedic1,2 or spinal cord injury.3 In dogs with osteoarthritis, regular swimming exercise for 8 weeks improves joint function, decreases pain, and reduces biochemical markers of joint disease.4 The weight of a swimming animal is supported by the water, which reduces compressive forces on joints and eliminates the work needed to overcome gravity, but the work needed to overcome drag is substantial. The energy required for human subjects to travel any distance when swimming is 4 times that when walking.5 Long-term swimming regimes prevent obesity in rats,6 and the integration of underwater and dry treadmill exercise into weight-loss programs for obese dogs preserves lean-body mass.7 For dogs, immersion in cold water increases energy expenditure markedly, and the loss of heat to water by conduction when swimming may also decrease the risk of exertional heatstroke in hot climates.8,9 Therefore, swimming has the potential to be a safe and effective component in the prevention and treatment of orthopedic disease and obesity.
Two studies10,11 have used video and motion-capture software to assess the limb kinematics of swimming dogs. In one of those studies,10 the range of motion (ROM) of the hip and stifle joints and tarsus were compared between healthy dogs and dogs recovering from surgery for cranial cruciate rupture while the dogs were walking and swimming with a flotation vest (FV) in a pool containing static water (ie, the dogs were not swimming against a current). Swimming elicited a greater ROM for the stifle joint (measured as the maximum change in joint angle from extension to flexion) than did walking.10 The other study11 evaluated the efficacy of various brands of canine FVs for awake and sedated dogs when exercised on an underwater treadmill with a water current of 5.6 km/h. The ROM of the elbow joint (measured as the change in distance between the humeral condyle and carpus) was greatest when dogs were swimming without an FV.11
Neither of those studies10,11 evaluated the effects of varying water flow rate (WFR) on limb kinematics and the rate of change of limb position or compared forelimb and hind limb kinematics concurrently. Frequency, excursion, and the product of excursion and frequency (ExF) of the carpus and tarsus should provide an indication of the rapidity and extent of limb movements in water. For dogs, when exercised on land, the speed of travel is the product of stride length and frequency, and both stride length and frequency increase with speed.12 However, when dogs are swimming, the respective relationships between WFR and frequency, excursion, and ExF are likely to be very different than when they are exercising on land because the limbs are interacting with a fluid medium rather than a solid surface.
The purpose of the study reported here was to evaluate how an FV and WFR affect the limb kinematics of dogs swimming against a current. The primary null hypothesis was that the kinematic parameters of the forelimbs and hind limbs do not change with WFR or the presence of an FV. Preliminary observations suggested the following subsidiary null hypotheses: that the swimming kinematics of dogs were not affected by fat mass or whether the WFR was incrementally increased or decreased. It was hoped that the study findings would help clarify whether WFR is an important factor during water-based exercises for obese dogs and dogs with orthopedic disease.
Materials and Methods
Animals
All study procedures were reviewed and approved by the University of Florida Institutional Animal Care and Use Committee (IACUC ID No. 201708224). A convenience sample of 7 healthy adult (age range, 3 to 10 years) Siberian Huskies (1 sexually intact male and 6 sexually intact females) from 1 breeder-owner were enrolled in the study with the owner’s consent. The dogs were selected because the owner was willing to bring the dogs to the laboratory 3 times/wk for many months and adjust the dogs’ food intake to maintain their body weight and condition. The 6 female dogs (S, Bl, By, K, A, and J; Supplementary Table S1) shared a common granddam or great granddam. Dog S was the dam of Bl, By, and K, each of which were from different litters. Dog Bl was the dam of A and J, which were littermates. All 7 dogs were orthopedically normal and considered healthy on the basis of results of a physical examination, CBC, serum biochemical analysis including determination of thyroid hormone concentrations, and urinalysis.
For each dog, the length of each limb was measured while the dog was standing. Each forelimb was measured from the ground to the most dorsal aspect of the scapula (withers), and each hind limb was measured from the ground to the femoral head. During each visit to the laboratory, each dog was individually weighed on a digital scale (Vet-Tec 300; Technidyne Scales Inc) and assigned a body condition score (BCS) on a validated scale13 of 1 to 9, where 1 is emaciated and 9 is morbidly obese. After each dog completed its training, total body waster was estimated by use of the deuterium oxide dilution method as described.14 Briefly, a venous blood sample (approx 2 mL) was obtained from each dog shortly before and 4 hours after administration of deuterium oxide (0.3 g/kg, SC) in physiologic saline (0.9% NaCl) solution. The dog did not have access to water for 2.5 hours before to 4 hours after the deuterium oxide injection. Deuterium enrichment in the blood was measured by use of mass spectrometry at a commercial laboratory (Metabolic Solutions Inc). The fat mass of each dog was calculated as the difference between body weight and fat-free mass, assuming that the total body water comprised 73.2% of the fat-free mass for each dog.15
Swimming habituation protocol
All dogs were habituated and trained to swim consistently against the current in a continuous-flow pool (SwimEx 600 S Series) over a period of 3 to 6 months by providing positive encouragement and minimal restraint as described.10 Throughout the study, each dog swam beside an investigator (CJF, HKR, MAR, or CMR) who stood in the pool and encouraged it to swim in a straight line. Investigators were trained to handle the dogs by use of fear-free methods (Fear Free LLC) and positive operant conditioning. All 7 dogs were show dogs and were accustomed to extensive human handling. They had also been previously trained to run for up to 30 minutes on a treadmill. A positive encouragement method was tailored to each dog to provide the strongest and most consistent motivation. It involved one or more of the following: feeding low-calorie treats, positive verbal and gentle tactile encouragement, and a clicking noise, which the dogs had been habituated to associate with a reward.16,17
Dogs were initially accustomed to the pool while wearing an FV (Kyjen 2518 Dog Life Jacket Quick Release) with no water flow until they were able to tread water calmly. Then, mild manual reassurance and support were used to train the dogs to swim without an FV. The reassurance and support were gradually reduced and the duration of each swim session and the WFR against which the dog had to swim were incrementally increased until each dog could swim comfortably for 30 min/d with or without an FV. The first 5 dogs enrolled in the study began each experimental session by swimming at the maximum WFR at which they were comfortable and then the WFR was gradually decreased to 0. Part way through the study, we recognized that dogs responded differently when WFR was either incrementally increased or incrementally decreased. Those first 5 dogs were subsequently trained to swim against gradually increasing WFRs. The last 2 dogs enrolled in the study were trained to swim against both gradually increasing and decreasing WFR concurrently.
The pool measured 3.6 X 1.8 X 1.2 m and the water had a mean pH of 7.2 and temperature of 28 °C. The pool had adjustable WFR settings that ranged from 0 to 99 units. Each dog swam at 7 different WFRs from 0 to 60 units, with each WFR differing by 10-unit increments. Each WFR was empirically measured in kilometers per hour by video recording the time required for a standard urine sample cup full of purple dye to float with consistent current speed for 3 m within the surface current from the end of the pool with the current-producing jets toward the other end of the pool. Pool WFRs settings of 0, 10, 20, 30, 40, 50, and 60 units correlated to empiric measurements of 0, 0.5, 1, 1.4, 1.9, 2,4, and 2.9 km/h, respectively. Those measured WFRs were slightly slower than the pool manufacturer’s swim current calculator (0, 0.6, 1.1, 1.7, 2.3, 2.8, and 3.4 km/h). Each dog swam for no more than 30 min/d 2 to 3 d/wk with at least 1 day of rest between swim days.
Study design
The study had a repeated-measures design by which each dog underwent each of 4 experimental sessions with at least 1 day between sessions. During each session, a dog swam with or without an FV against WFRs between 0 and 2.9 km/h that were incrementally decreased or increased. The right carpus and tarsus were wrapped in red-and-yellow self-adherent tape (Vetrap; 3M Animal Care Products) to facilitate visibility of paw location on video recordings. All dogs tolerated those markers well without habituation. One of 2 digital cameras (iPhone 6 plus; Apple Inc; Stylus Tough TG-tracker; Olympus Corp) were used to video record a dog’s leg movements through an underwater window at 240 frames/s. Leg movements were recorded as dogs swam for 1 to 2 minutes at each of 7 WFRs. Initially, leg movements were recorded at each WFR as dogs swam against a current, which was incrementally decreased from 2.9 to 0 km/h. Subsequently, video recordings were obtained at each WFR as the WFR was incrementally increased from 0 to 2.9 km/h because subjective observation and a preliminary trial involving 2 dogs suggested that the order of the WFRs to which dogs were exposed affected stroke kinematics.
Motion tracking software (Kinovea, version 0.8.15.Exe) was used to measure stroke excursion and frequency of the right forelimb and hind limb from video recorded at each WFR. Measurements were made during periods when the dog was swimming with consistent strokes in a straight line with minimal to no restraint. Stroke frequency (reported in strokes per minutes) was calculated by dividing the number of complete strokes during a period of approximately 15 seconds by the total time (in seconds) required to complete those strokes then multiplying by 60. Stroke excursion (reported in centimeters per stroke) was the distance the right carpus or tarsus traveled during a single stroke as measured by instructing the software to digitally track the tape marker on the leg (Figure 1). The mean for 10 excursion measurements and 2 separate 15-second frequency measurements were calculated for each WFR. The ExF (reported in meters per minute) represented the distance that each limb traveled through the water each minute. A 16 X 16-cm polyvinyl chloride (PVC) square with 5 X 5-cm markings was held adjacent to each dog during each video recording to provide a measure of scale (Supplementary Figure S1).10
Representative photographs of a healthy 4-year-old sexually intact female Siberian Husky that depict the excursion of the right carpus (A and C) and tarsus (B and D) when the dog was and was not wearing an FV and swimming against a current in a continuous-flow pool. The right carpus and right tarsus were wrapped with red-and-yellow self-adherent tape to facilitate visualization of paw location during video recordings. Motion tracking software was used to track the motion of the tape-markers and measure the stroke excursion and frequency of the right limbs. The pale pink line represents the track of the limb during complete excursion (A and B). The bold pink line represents how far the limb has progressed along the excursion track at that particular time (all panels). An investigator was standing beside the dog to provide it with encouragement and minimal restraint while it was swimming.
Citation: American Journal of Veterinary Research 82, 12; 10.2460/ajvr.21.02.0021
Representative photographs of a healthy 4-year-old sexually intact female Siberian Husky that depict the excursion of the right carpus (A and C) and tarsus (B and D) when the dog was and was not wearing an FV and swimming against a current in a continuous-flow pool. The right carpus and right tarsus were wrapped with red-and-yellow self-adherent tape to facilitate visualization of paw location during video recordings. Motion tracking software was used to track the motion of the tape-markers and measure the stroke excursion and frequency of the right limbs. The pale pink line represents the track of the limb during complete excursion (A and B). The bold pink line represents how far the limb has progressed along the excursion track at that particular time (all panels). An investigator was standing beside the dog to provide it with encouragement and minimal restraint while it was swimming.
Citation: American Journal of Veterinary Research 82, 12; 10.2460/ajvr.21.02.0021
Representative photographs of a healthy 4-year-old sexually intact female Siberian Husky that depict the excursion of the right carpus (A and C) and tarsus (B and D) when the dog was and was not wearing an FV and swimming against a current in a continuous-flow pool. The right carpus and right tarsus were wrapped with red-and-yellow self-adherent tape to facilitate visualization of paw location during video recordings. Motion tracking software was used to track the motion of the tape-markers and measure the stroke excursion and frequency of the right limbs. The pale pink line represents the track of the limb during complete excursion (A and B). The bold pink line represents how far the limb has progressed along the excursion track at that particular time (all panels). An investigator was standing beside the dog to provide it with encouragement and minimal restraint while it was swimming.
Citation: American Journal of Veterinary Research 82, 12; 10.2460/ajvr.21.02.0021
Dogs tended to move forward in the current when the WFR was slow and move backward when the WFR was fast. The amount of force necessary to prevent that tendency was measured by use of a spring scale (Model 8008-PN; Ohaus Corp) for 2 female dogs while they were swimming with and without an FV as WFR incrementally decreased. The investigator who was responsible for guiding the dogs in the pool held and monitored the spring scale, which was attached to the dog’s collar.
Statistical analysis
Descriptive data were generated, and results were summarized as the arithmetic mean ± SD or coefficient of variation (CV) or geometric mean (GM) ± geometric SD. The changes in kinetic parameters in response to changes (increase or decrease) in the WFR were calculated as a percentage of the values for the kinetic parameters when there was no current (WFR = 0). The Pearson correlation coefficient (R) was used to assess the correlation between fat mass and the mean excursion across all WFRs as the WFR incrementally decreased and when the dogs were (n = 7) and were not (6) wearing FVs. Trend lines were plotted by use of a spreadsheet program (Excel 2016; Microsoft Corp).
Prior to statistical comparisons, the excursion values underwent a logarithmic transformation because the variance for the forelimbs differed from that for the hind limbs. The transformed excursion values and untransformed values of frequency and total excursion (sum of ExF for the right forelimb and hind limb combined) were compared by use of generalized linear mixed models (Proc Glimmix; SAS for Windows 9.4; SAS Institute), with an autoregressive covariance structure used to account for correlation owing to repeated measures within dogs. Each model included a fixed effect for leg type (forelimb or hind limb), presence or absence of an FV, WFR, and WFR order (incrementally decreasing or increasing WFR), and all possible interactions and a random effect for dog. Variances were allowed to vary by leg and FV combination in the model for excursion and by the presence of an FV (yes or no) in the models for frequency and total excursion. For all models, values of P < 0.05 were considered significant.
Results
Dogs
For the 6 female dogs, body weight ranged from 19.3 to 24.7 kg, BCS ranged from 4.5 to 6, and fat mass ranged from 3.3 to 7.7 kg (Supplementary Table S1). The forelimb length ranged from 52 to 57 cm, and hind limb length ranged from 44 to 54 cm. Compared with the female dogs, the male dog was larger, taller, and leaner, with a body weight of 28.7 kg, BCS of 4.6, and negative fat mass, which was assumed to be 0. The male dog had a forelimb length of 59 cm and a hind limb length of 49 cm.
The male dog was unable to remain afloat when not wearing an FV unless the investigator held its tail to support its hindquarters. Thus, the variability in frequency and excursion of the forelimb for each individual as the WFR was incrementally decreased was reported for all 7 dogs when the FV was present but only the 6 female dogs when the FV was absent. The male dog and 1 female dog (S) had to be withdrawn from the study for reasons unrelated to the experimental procedures before video recordings could be obtained as the WFR was incrementally increased. Therefore, comparisons of the kinematic parameters under all conditions, including when WFR was both incrementally increased and decreased, were only reported for 5 female dogs.
Stroke variability
For all 7 dogs, the CVs for the mean excursion and frequency of the forelimb of each dog across all WFRs were 43 and 8%, respectively, when dogs were wearing an FV and swimming against a current that was incrementally decreasing (Figure 2). Leg length appeared to have little effect on the variation in excursion because the CV of mean excursion decreased by only 3% when excursion was normalized by dividing excursion by leg length. The CVs for the mean excursion and frequency across all WFRs were 21% and 11%, respectively, for the 6 female dogs when they were not wearing an FV and swimming against a current that was incrementally decreasing. When swimming against incrementally decreasing WFRs, the mean forelimb excursion across all WFRs was not correlated with fat mass regardless of whether the dogs were (R = –0.53; P = 0.20; n = 7) or were not (R = –0.17; P = 0.70; 6) wearing an FV.
Plots of stroke excursion (A) and frequency (B) for the right forelimbs of 7 healthy adult Siberian Huskies when the dogs were wearing FVs and swimming against a WFR that was incrementally decreased from 2.9 to 0 km/h. Results for each dog are represented by a different symbol, and the dotted lines represent the second-order polynomial fitted trend lines for each dog.
Citation: American Journal of Veterinary Research 82, 12; 10.2460/ajvr.21.02.0021
Plots of stroke excursion (A) and frequency (B) for the right forelimbs of 7 healthy adult Siberian Huskies when the dogs were wearing FVs and swimming against a WFR that was incrementally decreased from 2.9 to 0 km/h. Results for each dog are represented by a different symbol, and the dotted lines represent the second-order polynomial fitted trend lines for each dog.
Citation: American Journal of Veterinary Research 82, 12; 10.2460/ajvr.21.02.0021
Plots of stroke excursion (A) and frequency (B) for the right forelimbs of 7 healthy adult Siberian Huskies when the dogs were wearing FVs and swimming against a WFR that was incrementally decreased from 2.9 to 0 km/h. Results for each dog are represented by a different symbol, and the dotted lines represent the second-order polynomial fitted trend lines for each dog.
Citation: American Journal of Veterinary Research 82, 12; 10.2460/ajvr.21.02.0021
The mean force required to prevent the 2 evaluated dogs from moving backward in the pool when swimming against a current was small (< 0.6 kg at all WFRs), was lower when the dog was wearing an FV versus when it was not wearing an FV, and decreased as the WFR decreased regardless of whether the dog was or was not wearing an FV. That force became slightly negative at low WFRs when the dogs were not wearing FV, which indicated a slight application of force was necessary to prevent the dogs from moving forward in the pool (Supplementary Figure S2).
Stroke frequency
For the 5 female dogs that completed the experimental protocol, the mean stroke frequency across all experimental conditions was 64 strokes/min (range, 40 to 90 strokes/min; Figure 3). The mean stroke frequency under any condition did not differ significantly between the forelimb and hind limb and there was no evidence of a significant interaction between leg type and any of the other fixed effects (P > 0.70).
Mean stroke frequency (A) and excursion (B) for the right forelimb (circles) and hind limbs (triangles) of 5 healthy adult sexually intact female Siberian Huskies when they were (black symbols) and were not (white symbols) wearing an FV and were swimming against a current with an incrementally increasing (solid lines) or decreasing (dashed lines) WFR. Error bars represent the SEM.
Citation: American Journal of Veterinary Research 82, 12; 10.2460/ajvr.21.02.0021
Mean stroke frequency (A) and excursion (B) for the right forelimb (circles) and hind limbs (triangles) of 5 healthy adult sexually intact female Siberian Huskies when they were (black symbols) and were not (white symbols) wearing an FV and were swimming against a current with an incrementally increasing (solid lines) or decreasing (dashed lines) WFR. Error bars represent the SEM.
Citation: American Journal of Veterinary Research 82, 12; 10.2460/ajvr.21.02.0021
Mean stroke frequency (A) and excursion (B) for the right forelimb (circles) and hind limbs (triangles) of 5 healthy adult sexually intact female Siberian Huskies when they were (black symbols) and were not (white symbols) wearing an FV and were swimming against a current with an incrementally increasing (solid lines) or decreasing (dashed lines) WFR. Error bars represent the SEM.
Citation: American Journal of Veterinary Research 82, 12; 10.2460/ajvr.21.02.0021
Mean stroke frequency (A) and excursion (B) for the right forelimb (circles) and hind limbs (triangles) of 5 healthy adult sexually intact female Siberian Huskies when they were (black symbols) and were not (white symbols) wearing an FV and were swimming against a current with an incrementally increasing (solid lines) or decreasing (dashed lines) WFR. Error bars represent the SEM.
Citation: American Journal of Veterinary Research 82, 12; 10.2460/ajvr.21.02.0021
Mean stroke frequency (A) and excursion (B) for the right forelimb (circles) and hind limbs (triangles) of 5 healthy adult sexually intact female Siberian Huskies when they were (black symbols) and were not (white symbols) wearing an FV and were swimming against a current with an incrementally increasing (solid lines) or decreasing (dashed lines) WFR. Error bars represent the SEM.
Citation: American Journal of Veterinary Research 82, 12; 10.2460/ajvr.21.02.0021
The mean stroke frequency across all conditions was 17% greater when the dogs were not wearing an FV (69 strokes/min), compared with when the dogs were wearing an FV (59 strokes/min; P < 0.001). Additionally, mean stroke frequency was significantly (P = 0.005) affected by the interaction between the presence of an FV and WFR order. When the dogs were not wearing an FV, the mean stroke frequency increased by 15 strokes/min when the WFR was incrementally decreased but increased by only 4 strokes/min when the WFR was incrementally increased, compared with when the dogs were wearing an FV.
The mean stroke frequency across all conditions increased with WFR (P < 0.001) but mean stroke frequency was significantly (P = 0.006) affected by the interaction between WFR and WFR order. The mean stroke frequency changed little with WFR when the WFR was incrementally decreased but increased by 11% with WFR when the WFR was incrementally increased. In addition, mean stroke frequency was significantly (P = 0.01) affected by the interaction among the presence of an FV, WFR, and WFR order. The mean stroke frequency increased by 5 and 9 strokes/min as the WFR was incrementally increased when the dogs were and were not wearing an FV, respectively.
Excursion
For the 5 female dogs that completed the experimental protocol, the GM excursion of the hind limb (15 cm/stroke) was only 30% that of the forelimb (50 cm/stroke; P < 0.001) across all conditions (Figure 3). The shape of the excursion path also differed between the carpus and tarsus. The carpus followed an elongated oval-shaped path, whereas the tarsus followed a more condensed circular excursion path (Figure 1). There was no evidence that excursion was significantly affected by any interactions between leg type and any of the other fixed effects.
Excursion was significantly (P < 0.001) affected by the presence of an FV. When dogs were wearing an FV, the GM excursion across all conditions decreased by 38% compared with when they were not wearing an FV (a 43% and 33% decrease of GM excursion in the hind limb and the forelimb, respectively). The GM excursion increased significantly (P < 0.001) as the WFR increased from 0 to 2.9 km/h. The GM excursion was also significantly (P = 0.002) affected by the interaction between WFR and WFR order. The GM excursion increased by 69% with WFR when the WFR was incrementally increased from 0 to 2.9 km/h but was increased by only 32% with WFR when the WFR was incrementally decreased from 2.9 to 0 km/h.
Total excursion of both limbs combined
For the 5 female dogs that completed the experimental protocol, the mean total excursion of both legs combined across all conditions decreased 42% from 5.9 m/min when the dogs were not wearing an FV to 3.4 m/min when the dogs were wearing an FV (P < 0.001; Figure 4). Mean total excursion increased significantly (P < 0.001) as WFR increased. Total excursion was also significantly (P < 0.001) affected by the interaction between WFR and WFR order. The mean total excursion increased 69% with WFR when the WFR was incrementally increased but increased by only 19% with WFR when the WFR was incrementally decreased.
Mean total excursion (ie, sum of the ExF for the right forelimb and hind limb combined) for the dogs of Figure 3 when they were (black symbols) and were not (white symbols) wearing an FV and were swimming against a current with an incrementally increasing (solid lines) or decreasing (dashed line) WFR. Error bars represent the SEM.
Citation: American Journal of Veterinary Research 82, 12; 10.2460/ajvr.21.02.0021
Mean total excursion (ie, sum of the ExF for the right forelimb and hind limb combined) for the dogs of Figure 3 when they were (black symbols) and were not (white symbols) wearing an FV and were swimming against a current with an incrementally increasing (solid lines) or decreasing (dashed line) WFR. Error bars represent the SEM.
Citation: American Journal of Veterinary Research 82, 12; 10.2460/ajvr.21.02.0021
Mean total excursion (ie, sum of the ExF for the right forelimb and hind limb combined) for the dogs of Figure 3 when they were (black symbols) and were not (white symbols) wearing an FV and were swimming against a current with an incrementally increasing (solid lines) or decreasing (dashed line) WFR. Error bars represent the SEM.
Citation: American Journal of Veterinary Research 82, 12; 10.2460/ajvr.21.02.0021
Discussion
Results of the present study indicated that, for healthy adult dogs evaluated while swimming in a continuous-flow pool, all assessed kinematic variables (stroke excursion, frequency, and ExF) increased with WFR when the WFR was incrementally increased from 0 to 2.9 km/h. That finding was similar to the response to an increasing WFR observed when human subjects,18 muskrats,19 ducks,20 and mink21 are swimming. Surprisingly, the kinematic parameters for the dogs of the present study changed little with WFR when the WFR was incrementally decreased from 2.9 to 0 km/h. This suggested that the dogs were less responsive to changes in WFR when it was gradually decreased versus when it was gradually increased, perhaps because the adrenaline release in response to the intense activity required to swim against a faster water current blunted the dogs’ ability to recognize that they did not have to work as hard against a slower current. It is also possible that differences in the pool habituation protocols influenced the results of the present study. Only the last 2 dogs enrolled in the study were exposed to increasing and decreasing WFRs from the beginning of their acclimation training. The other dogs were only habituated to swim against a gradually decreasing WFR because we did not recognize the significant effect of WFR order on the kinematic parameters until part way through the study.
For the dogs of the present study, limb excursion increased by 49% but stroke frequency increased by only 5% as the WFR increased from 0 to 2.9 km/h. Additionally, stroke excursion differed substantially, whereas F was almost identical among individual dogs when the WFR was incrementally decreased. This suggested that swimming dogs primarily adjust their limb excursion not stroke frequency when responding to changes in current, which was consistent with the findings for other species when swimming.18–21 It also suggested that slowly increasing ExF by incrementally increasing WFR may be a useful strategy for physical rehabilitation of dogs. Increasing the ROM of the limbs increases blood flow, stimulates stretch receptors, reduces contraction, and promotes recovery from various orthopedic and neurologic conditions including osteoarthritis, cranial cruciate ligament rupture, fibrocartilaginous embolism, degenerative myelopathy, and intervertebral disc disease.22–27
Limb excursion differed substantially among the dogs of the present study even though they were similar in size and many were closely related. The male dog had the longest legs and the largest excursion and ExF but normalizing the kinematic parameters on the basis of leg length did not substantially decrease the extent of variation in excursion among dogs. The large variation in limb excursion among dogs suggested that some dogs were more efficient swimmers than others despite the fact that all dogs underwent similar training. The male dog had almost no fat mass and sank when swimming without an FV unless its hind quarters were supported, which made us question whether the inter-subject variation in excursion might be attributable to differences in the fat mass of individual dogs, which affected buoyancy. However, a strong correlation between limb excursion and fat mass was not identified when dogs were swimming without an FV. An inverse relationship between buoyancy and total stroke kinematics has been reported in terrestrial and semiaquatic opossums; the non-wettable fur of semiaquatic opossums increases buoyancy sufficiently to allow them to paddle with only their hind limbs.28 Athletic human swimmers have significantly higher body fat percentages than other competitive athletes29,30; however, swimming speed decreases when the fat mass of athletic swimmers is artificially increased.31 There may be an ideal amount of fat or distribution of fat that provides increased buoyancy without increasing drag in dogs as well as human swimmers. This relationship needs further evaluation, especially in retrieving breeds that are bred to swim without an FV and among dogs with a large fat mass range.
In the present study, the stroke excursion, frequency, and ExF for both the forelimb and hind limb decreased when dogs were swimming while wearing an FV, compared with when they were not wearing an FV, which was consistent with the results of another study11 in which dogs were evaluated while swimming. In that study,11 the ROM for the elbow joint decreased between 4% and 33% when the dog was wearing an FV (depending on the type of FV). In the present study, the forelimb excursion decreased 38% when dogs were wearing an FV relative to when they were not wearing an FV. A direct comparison between that other study11 and the present study was not possible because, in the other study, a different measure of limb motion was evaluated in a different population of dogs, which were not habituated to the experimental protocol prior to its initiation, and dogs were evaluated while wearing different types of FVs and exercising on an underwater treadmill at different WFRs than were used in the present study. The investigators of the other study11 postulated that the differences in the elbow joint ROM among types of FVs were associated, in part, with a change in the swimming position of the dogs (ie, from the rump being below the water surface when dogs were not wearing an FV to the rump being raised above the water surface to varying extents when the various FVs were worn). All 6 female dogs of the present study swam with their rumps close to the water surface and no obvious difference in swimming position was appreciated regardless of whether they were or were not wearing an FV. The water current speed in the other study11 (5.6 km/h) was greater than any WFR that the dogs of the present study could swim against for any sustained period of time. However, the method by which the speed of the water current was determined was not reported in the other study.11 The empirical speed of the WFR in the pool used in the present study was less than that specified by the manufacturer, and the same might have been true for the pool used in the other study.11
Flotation vests provide buoyancy, which eliminates the need to combat gravity and allows limb movements to be used solely to oppose water flow. In the present study, when the dogs were not wearing an FV, the increase in limb movement to combat gravity was achieved by increasing excursion rather than stroke frequency, which suggests that swimming protocols for physical rehabilitation should involve dogs wearing an FV initially and that swimming without the aid of an FV may be beneficial during the latter stages of rehabilitation and for weight loss. The force exerted on the investigator by the swimming dogs of the present study was small and more negative when dogs were wearing versus were not wearing an FV. This indicated that, in the absence of an FV, the more rigorous limb movements required to maintain buoyancy contribute to some forward motion through the water.
To our knowledge, concurrent comparison of forelimb and hind limb stroke kinetics in swimming dogs had not been performed prior to the present study. The frequency and length of each stride are, by necessity, the same for both the forelimbs and hind limbs when dogs traverse solid surfaces,32 but those constraints do not necessarily apply when dogs are swimming. Interestingly, for the dogs of the present study, the stroke frequency of the forelimb was similar to that of the hind limb, whereas the excursion of the forelimb was 3-fold greater than that of the hind limb across all conditions. Extension and flexion were more dramatic in the forelimb than in the hind limb, especially at low WFRs. In human subjects and other animals, such as otters, the relative contribution of the thoracic limb (forelimb) and pelvic limb (hind limb) to forward motion when swimming varies depending on species and type of stroke.33 It remains to be determined whether dogs that are bred for swimming (eg, retrievers) use different limb movement strategies when swimming than do dogs that are not bred for swimming.
The present study had several limitations. The study population consisted of a small number (n = 7) of closely related dogs of a single breed (Siberian Husky), all but 1 of which were sexually intact females. The age range for the dogs was fairly wide (3 to 10 years), and we did not specifically evaluate how age may have affected the measured variables. All 7 dogs were show dogs and accustomed to interactions with humans. Siberian Huskies have been bred primarily to pull sleds rather than swim in water. The study dogs were trained and habituated to the experimental protocol for several months, and the limb kinematics and responses to changes in WFR may differ for untrained pet dogs. Although the dogs of this study were acclimated to swim against the current, there was a tendency for them to move away from the center of the pool where the current was the strongest. Consequently, an investigator was stationed in the pool next to each dog during each experimental session. The investigator used minimal guidance to keep the dog swimming in a straight line in the center of the pool. The force required for that guidance helped the dogs move slightly forward against high WFRs and prevented the dogs from moving forward at low WFRs. The guidance force applied by the investigator was measured for 2 of the 7 study dogs and, although it was small, it could slightly diminish changes in swim kinetics in response to changes in WFR.10 That force should be measured under all conditions in all study subjects in future studies. Finally, the methods used to measure stroke kinematics in this study failed to account for the 3-D aspects of limb motion in water or the ROM of specific joints. A PVC square with 5-cm markings was used to eliminate distortion due to parallax, but future studies will benefit from direct linear transformation of coordinate reconstruction to correct for distortion within the calibrated volume of water. Instrumentation of dogs with > 1 digital tracking marker would also help assess how the ROM for specific joints changes with WFR.
Further studies involving more dogs than the present study and dogs of various breeds and with varying body conditions are necessary to determine whether the observations for the dogs of this study are representative for all dogs and to evaluate how buoyancy affects the swimming kinematics of dogs. We hypothesize that dogs bred for swimming will expend less energy when swimming than dogs not bred for swimming and that the increased adipose tissue in the hindquarters of female dogs relative to male dogs may provide buoyancy that minimizes the energy necessary for swimming. Findings of such studies could aid clinicians in designing physical rehabilitation and weight loss programs for dogs.
Findings of the present study indicated that it was possible to quantify and compare stroke kinematics among dogs swimming against a current in a continuous-flow pool. Results also indicated that Siberian Huskies use primarily their forelimbs to swim against a current at low WFRs, and only use their hind limbs substantially at higher WFRs. This implied that, for dogs, swimming may not be as useful for rehabilitation of hind limb injuries as it is for forelimb injuries except at high WFRs. For the dogs of the present study, stroke excursion, frequency, and ExF all increased with WFR when WFR was incrementally increased but not when WFR was incrementally decreased. Thus, requiring dogs to swim against incremental increases in WFR may be a better strategy for rehabilitation purposes than requiring dogs to swim against incremental decreases in WFR, and this warrants investigation in future studies. Results of this study confirmed previous observations11 that the ROM for the forelimb was greater when dogs were not wearing versus when they were wearing an FV. Stroke excursion, but not stroke frequency, differed markedly among the dogs of this study, and results suggested that dogs, like other species alter stroke excursion much more than stroke frequency to adjust speed through the water and compensate for differences in buoyancy. Collectively, these findings suggested that swimming protocols for rehabilitation purposes should initially begin with dogs wearing an FV and swimming against a low WFR and progress to the dogs swimming against higher WFRs without an FV to achieve maximum ROM for the limbs. The WFR should be gradually increased rather than decreased during each swimming session to maximize limb movements. In dogs, swimming has the potential to increase the ROM for the forelimbs to a greater extent than the ROM for the hind limbs.
Supplementary Materials
Supplementary materials are posted online at the journal website: avmajournals.avma.org
Acknowledgments
Supported by a donation from Dr. William Anspach in memory of his dog, Maggie.
The authors declare that there were no conflicts of interest.
The authors thank Dr. Aulus Carciofi for helping habituate the dogs to swimming and troubleshooting the initial phases of this experiment and Sally A. O’Connell for volunteering her dogs for this study.
References
- 1. ↑
Levine D, Marcellin-Little DJ, Millis DL, Tragauer V, Osborne JA. Effects of partial immersion in water on vertical ground reaction forces and weight distribution in dogs. Am J Vet Res. 2010;71(12):1413–1416.
- 2. ↑
Marsolais GS, Dvorak G, Conzemius MG. Effects of postoperative rehabilitation on limb function after cranial cruciate ligament repair in dogs. J Am Vet Med Assoc. 2002;220(9):1325–1330.
- 3. ↑
Gandini G, Cizinauskas S, Lang J, Fatzer R, Jaggy A. Fibrocartilaginous embolism in 75 dogs: clinical findings and factors influencing the recovery rate. J Small Anim Pract. 2003;44(2):76–80.
- 4. ↑
Nganvongpanit K, Tanvisut S, Yano T, Kongtawelert P. Effect of swimming on clinical functional parameters and serum biomarkers in healthy and osteoarthritic dogs. ISRN Vet Sci. 2014;2014:459809. doi:10.1155/2014/459809
- 5. ↑
McArdle WD, Katch FI, Katch VL. Energy expenditure during walking, jogging, running and swimming. In: McArdle WD, Katch FI, Katch VL, eds. Exercise Physiology—Energy, Nutrition, and Human Performance. 4th ed. Williams & Wilkins; 1996:167–186.
- 6. ↑
Melton SA, Hegsted M, Keenan MJ, et al. Swimming eliminates the weight gain and abdominal fat associated with ovariectomy in the retired breeder rat despite high-fat diet selection. Appetite. 2000;35(1):1–7.
- 7. ↑
Vitger AD, Stallknecht BM, Nielsen DH, Bjornvad CR. Integration of a physical training program in a weight loss plan for overweight pet dogs. J Am Vet Med Assoc. 2016;248(2):174–182.
- 8. ↑
National Research Council. Physical activity and environment. In: National Research Council. Nutrient Requirements of Dogs and Cats. National Academies Press; 2006:258–312.
- 9. ↑
Hall EJ, Carter AJ, O'Neill DG. Incidence and risk factors for heat-related illness (heatstroke) in UK dogs under primary veterinary care in 2016. Sci Rep. 2020;10(1):9128. doe:10.1038/s41598-020-66015-8
- 10. ↑
Marsolais GS, McLean S, Derrick T, Conzemius MG. Kinematic analysis of the hind limb during swimming and walking in healthy dogs and dogs with surgically corrected cranial cruciate ligament rupture. J Am Vet Med Assoc. 2003;222(6):739–743.
- 11. ↑
Corum CP, Wichtowski M, Hetts S, Estep D, Bertone JJ. Swimming kinematic and flotation analysis of conscious and sedated dogs using 3 canine flotation devices. Top Companion Anim Med. 2014;29(4):102–108.
- 12. ↑
Bertram JEA, Lee DV, Case HN, Todhunter RJ. Comparison of the trotting gaits of Labrador Retrievers and Greyhounds. Am J Vet Res. 2000;61(7):832–838.
- 13. ↑
Laflamme D. Development and validation of a body condition score system for dogs. Canine Pract. 1997;22(4):10–15.
- 14. ↑
Hill RC, Lewis DD, Scott KC, et al. Effect of increased dietary protein and decreased dietary carbohydrate on performance and body composition in racing Greyhounds. Am J Vet Res. 2001;62(3):440–447.
- 15. ↑
Pace N, Kline L, Howard K, et al. Studies on body composition; use of radioactive hydrogen for measurement in vivo of total body water. J Biol Chem. 1947;168(2):459–469.
- 16. ↑
Kelleher RT, Gollub LR. A review of positive conditioned reinforcement. J Exp Anal Behav. 1962;5(suppl 4):543–597.
- 17. ↑
Smith SM, Davis ES. Clicker increases resistance to extinction but does not decrease training time of a simple operant task in domestic dogs (Canis familiaris). Appl Anim Behav Sci. 2008;110(3–4):318–329.
- 18. ↑
Nadel ER. Thermal and energetic exchanges during swimming. In: Nadel ER, ed. Problems With Temperature Regulation During Exercise. Academic Press Inc; 1977:91–120.
- 19. ↑
Fish FE. Mechanics, power output and efficiency of the swimming muskrat (Ondatra zibethicus). J Exp Biol. 1984;110:183–201.
- 20. ↑
Prange HD, Schmidt-Nielsen K. The metabolic cost of swimming in ducks. J Exp Biol. 1970;53(3):763–777.
- 21. ↑
Williams TM. Locomotion in the North American mink, a semi-aquatic mammal. I. Swimming energetics and body drag. J Exp Biol. 1983;103:155–168.
- 22. ↑
Sapega AA, Quedenfeld TC, Moyer RA, Butler RA. Biophysical factors in range-of-motion exercise. Phys Sportsmed. 1981;9(12):57–65.
- 23.
Saunders DG, Walker JR, Levine D. Joint mobilization. Vet Clin North Am Small Anim Pract. 2005;35(6):1287–1316, vii-viii.
- 24.
Hamilton S. Therapeutic exercises. In: Millis D L, Levine D, eds. Canine Rehabilitation and Physical Therapy. WB Saunders; 2004:244–263.
- 25.
Olby N, Halling KB, Glick TR. Rehabilitation for the neurologic patient. Vet Clin North Am Small Anim Pract. 2005;35(6):1389–1409.
- 26.
Johnston SA, McLaughlin RM, Budsberg SC. Nonsurgical management of osteoarthritis in dogs. Vet Clin North Am Small Anim Pract. 2008;38(6):1449–1470, viii.
- 27. ↑
Au KK, Gordon-Evans WJ, Dunning D, et al. Comparison of short- and long-term function and radiographic osteoarthrosis in dogs after postoperative physical rehabilitation and tibial plateau leveling osteotomy or lateral fabellar suture stabilization. Vet Surg. 2010;39(2):173–180.
- 28. ↑
Fish FE. Comparison of swimming kinematics between terrestrial and semiaquatic opossums. J Mammal. 1993;74(2):275–284.
- 29. ↑
Jang KT, Flynn MG, Costill DL, et al. Energy balance in competitive swimmers and runners. J Swim Res. 1987;3:19–23.
- 31. ↑
Lowensteyn I, Signorile JF, Giltz K. The effect of varying body composition on swimming performance. J Strength Cond Res. 1994;8:149–154.
- 32. ↑
Maes LD, Herbin M, Hackert R, Bels VL, Abourachid A. Steady locomotion in dogs: temporal and associated spatial coordination patterns and the effect of speed. J Exp Biol. 2008;211(Pt 1):138–149.
- 33. ↑
Fish FE. Association of propulsive swimming mode with behavior in river otters (Lutra canadensis). J Mammal. 1994;75(4):989–997.
