In cyclical forms of locomotion, average speed
is the product of the average length of distance covered in each cycle
and the average number of cycles completed in a prescribed time
When running and walking, a person chooses a combination of stride length and stride rate that minimizes oxygen uptake (V˙O2) at a given speed (1,4-6,13,21). Deviation from this optimal combination of stride length and stride rate results in a marked increase in V˙O2, suggesting that minimizing energy cost may be an important factor contributing to cadence determination in cyclical forms of locomotion.
Like running, a similar U-shaped response of V˙O2 has been observed for variations in cycling cadence. Trained cyclists typically demonstrate minimum oxygen cost at a cadence of approximately 60 rpm, but unlike running, cyclists prefer to choose a cadence that is approximately 50% greater (90 rpm) than the cadence at which V˙O2 is minimized, perhaps to minimize muscle stress (10). Thus, the choice of cycle rate is not simply determined by minimizing energy cost.
Given the U-shaped response of V˙O2 to changes in cycle rate in constant speed gait and cycling, it is reasonable to expect such a relationship for swimming as another cyclical form of locomotion. However, data to support such a claim are limited. Using five stroke rate (SR) conditions ranging from 0.68 to 0.87 strokes per second, Swaine and Reilly (15) reported that average swimming speed was maximized at the median SR of 0.77 strokes per second. Furthermore, when these SR were used during simulated maximal effort swimming on a swim bench ergometer, V˙O2 was also maximized at the median SR (15). Yet, Pawelczyk et al. (12) presented preliminary data, indicating that when swimming speed was held constant, variations in stroke length (SL) up to ±10% of freely chosen SL at a swimming speed of 1.0 m·s−1 in flume swimming trials resulted in only small variations in HR and V˙O2.
Considering the mechanical differences between swimming and other forms of cyclical locomotion, the limited study of SR variation on swimming economy, and the equivocal results of these studies, it is not clear what relationship exists between oxygen uptake and SL or SR at a constant speed in swimming. However, such information would prove useful for understanding human movement optimization as well as for providing data upon which decisions about choosing and/or training stroke characteristics can be based. Therefore, the purpose of this project was to determine the effect of SL and SR manipulation on the physiological response when swimming at a constant submaximal speed. It was hypothesized that deviation from preferred SR will result in increased V˙O2 and HR and that this deviation will be a U-shaped response.
Six male and four female experienced middle-distance freestyle swimmers currently participating in a masters or collegiate swimming program (mean ± SD: age = 33.3 ± 13.6 yr, height = 175.3 ± 8.6 cm, weight = 74.9 ± 12.2 kg) volunteered to participate in the study. The data reported are representative of the nine swimmers who completed all stages of data collection. All participants provided informed consent to participate as approved by the institutional review board of Southwestern University.
Participants completed all swimming trials in a flume (Unidyne, Minneapolis, MN) that provided a 2.4 × 4.7-m area of flow that varied in depth from 0.6 m at the sides to 1.2 m in the center of the channel. Flow speed in the flume was measured at a depth of 1 ft below the water surface (±0.03 m·s−1) using an FP101 flow probe (Global Water, Gold River, CA). All trials were completed at a speed of 1.0 m·s−1. This speed was chosen on the basis of pilot data to accommodate the demanding requirements of the −20% SR condition (i.e., the longest SL). SR was manipulated using an aquatic metronome (Finis Tempo Trainer, Finis USA, Livermore, CA) that emitted an audible tone at a programmed frequency. This device was placed under a swim cap near the swimmer's temple. The metronome frequency was twice the target SR such that the swimmer was asked to match alternating hand entries to the frequency.
Participants first completed a 5-min accommodation freestyle swim in the flume without the use of the metronome, hereafter called the free swimming trial. Stroke counts made for 30-s intervals during each of the last 3 min of this swim were averaged and used to compute preferred SR.
Participants were then fitted with a specially designed respiratory valve that fixed the inspiratory and expiratory tubes vertically parallel. The valves in the inspiratory and expiratory tubing were placed in an extension of the mouthpiece to ensure a minimal "dead space" of 30 mL (16,17). The inspiratory tube (length = 1.65 m, diameter = 36 mm) was open to atmospheric air. Expired gasses passed through 0.74 m of 36-mm-diameter tubing to the gas analysis system. Participants were then accommodated to swimming with this device in place by completing a minimum of 5 min of freestyle swimming and continued until the participants reported that they felt comfortable swimming with the respiratory valve in place.
Participants then completed a series of five freestyle swims during which oxygen uptake was monitored. Five SR conditions, 0%, ±10%, and ±20% of preferred SR, were randomly presented. The participants were not informed of the SR but were instructed to simply match their SR to the frequency of the metronome. Each swim continued for 1 min after steady-state oxygen uptake was established (∼4-5 min). Immediately after completion of each trial, participants indicated their RPE using the Borg scale. Participants were given a rest period that lasted no longer than 4 min during which the metronome was removed, reprogrammed, and replaced on the swimmer.
Oxygen uptake was measured simultaneously with the Douglas bag technique and an online system for breath-by-breath measurements. Douglas bags were considered standard for all V˙O2 measurements. Breath-by-breath data served as backup and were used for identification of steady states. The online system for breath-by-breath measurements consisted of four one-way valves to direct flow, two sample lines to measure gas fractions, and a turbine flow meter (VMM, Interface Associates, Aliso Viejo, CA) to measure ventilation. Breath-by-breath data were stored on a computer and analyzed using customized software. The Douglas bag gas fractions were analyzed by a mass spectrometer (Marquette MGA 1100, St. Louis, MO) calibrated before each test. Ventilatory volume was measured with a Tissot spirometer. HR was recorded using a Polar S610i HR monitor (Polar USA, Lake Success, NY) and was averaged across the final minute of data collection to correspond to steady-state V˙O2.
SR was monitored during each minute of each trial by measuring the time needed to complete 10 stroke cycles. These values were compared with target values to ensure that the swimmer maintained the appropriate SR. A video camera operating at 60 Hz was positioned to record underwater stroke movements through an underwater window located in the flume. From the video recording, kick rate (KR) was determined by measuring the time needed to complete 60 kicks (30 kick cycles) during the final minute of each trial.
A one-way repeated-measures analysis of variance with a Tukey post hoc analysis was used to evaluate changes in V˙O2 as a function of SR, with V˙O2 as the repeated dependent variable and five levels of SR as the independent variable. Partial η2 and Rosenthal's (14) effect size (ES) were reported to aid interpretation of mean comparisons. Similar repeated-measures ANOVA were used to evaluate RPE, HR, and KR data as a function of SR.
When using a metronome to control SR, swimmers were able to match SR to the predetermined frequency across all conditions (Table 1). Furthermore, on the basis of the results of a dependent t-test, preferred SR, as measured during the free swimming trial (i.e., no metronome), was not different from the measured SR during the 0% preferred SR trial (i.e., with metronome) (t(8) = 1.01, P = 0.341). Differences between the measured SR for the five SR conditions were verified using a one-way repeated-measures ANOVA (F(4, 32) = 1146.2, P < 0.001, partial η2 = 0.993) with pairwise comparisons, indicating that all SR were significantly different (P < 0.001) from each other. KR was significantly affected by altering SR from the preferred cadence (F(4, 32) = 10.371, P < 0.001, partial η2 = 0.565). Specifically, when SR was decreased by 10%, KR increased by 21% (P = 0.107, ES = 0.72) compared with the KR measured at the preferred cadence and by 36% (P < 0.001, ES = 0.95) when SR was decreased by 20%. However, when SR was increased above the preferred cadence, KR increased by a modest 5%-8%, which was not significantly increased relative to that measured at the preferred SR.
Oxygen uptake was significantly (F(4, 32) = 12.012, P = 0.002, partial η2 = 0.60) affected by altering SR from the preferred cadence (Fig. 1). Specifically, V˙O2 increased by 11.4% (P = 0.015, ES = 0.73) over the V˙O2 at the preferred cadence when SR was decreased by 10% and by 16.4% (P < 0.001, ES = 1.11) when SR was decreased by 20%. However, when SR was increased above the preferred cadence, V˙O2 was not significantly increased relative to that measured at the preferred SR (ES = 0.22 and 0.03 at +10% and +20% SR, respectively).
Similarly, HR was significantly (F(4, 32) = 8.728, P < 0.001, partial η2 = 0.52) affected by altering SR from the preferred cadence (Fig. 1). When SR was decreased by 10%, HR increased by 4.0% (P = 0.039, ES = 0.31) compared with the HR measured at the preferred cadence and by 6.1% (P = 0.001, ES = 0.46) when SR was decreased by 20%. However, when SR was increased above the preferred cadence, HR was not significantly increased relative to that measured at the preferred SR (ES = 0.04 and 0.15 at +10% and +20% SR, respectively).
RPE was significantly (F(4, 32) = 9.466, P < 0.001, partial η2 = 0.54) affected by altering SR from the preferred cadence (Fig. 2). Specifically, when SR was decreased by 10%, RPE increased by 15.4% (P = 0.106, ES = 0.91) compared with the RPE measured at the preferred cadence and by 29.8% (P < 0.001, ES = 1.44) when SR was decreased by 20%. However, when SR was increased above the preferred cadence, RPE was not significantly increased relative to that measured at the preferred SR (ES = 0.16 and 0.41 at +10% and +20% SR, respectively).
Numerous studies have examined the manipulation of cycle length and rate on physiological demands but only for activities that have little, if any, upper-body contribution to movement speed. The present study demonstrated that in swimming, an activity that derives substantial propulsion from the arms, alterations in SR accounted for more than 60% of the variance in V˙O2 and 52% of the variance in HR. Although visual inspection of these data suggested a U-shaped response across SR conditions, when these data were fit to a quadratic equation, the coefficient of the quadratic term in these fits was not significantly different from zero (P = 0.17 and 0.43, respectively). This suggested that in this sample of participants, these physiological responses did not follow the U-shaped pattern observed in other forms of locomotion.
Although alterations in SR did not produce U-shaped responses in V˙O2 and HR, the response to reducing SR (i.e., lengthening SL) was an increased V˙O2 and HR, which was similar to what has been observed in leg-dominated cyclical activities. When walking, a 5% deviation in stride rate resulted in an 8%-10% increase in V˙O2 (6), whereas a 10% reduction in SR achieved a similar 11% increase in V˙O2. However, an increased SR did not substantially increase V˙O2 or HR when swimming. This response was unlike that reported for arm cranking, another form of exercise dependent on the upper extremity, where aerobic demand was minimized at lower cadences (13) and preferred cadences were higher than the most economical cadence (19).
Unlike gait or cycling, swimming is a movement that relies heavily on the interaction of upper- and lower-body movements (20). Propulsion in swimming is generated from the actions of the arms and the legs. Alterations in these propulsive components across conditions may have affected V˙O2. Swimming with a lower SR was accompanied by a substantial increase in KR. This increase in KR paralleled the increase in V˙O2, suggesting that a greater degree of involvement of the muscle mass in the legs likely accounted for the increased oxygen uptake (Fig. 3).
As a whole body activity, swimming may not be characterized by the typical U-shaped response in V˙O2 and HR because the contributions from the upper- and lower-body change disproportionately to meet the demands of the conditions. At lower SR, higher KR are used to maintain speed and vice versa. Given that V˙O2 remains relatively stable at higher SR, these data suggest that swimmers choose an SR that keeps V˙O2 low but with the longest SL. This is unlike cyclists who choose a cadence higher than the most economical or runners who choose a stride rate that minimizes V˙O2 (10).
Contrary to the present study, increasing cycle rate in constant power arm-cranking exercise increased submaximal oxygen consumption (8,9). It is possible that the physiological response may have been different if the movement was constrained to the arms only, but such a situation would not adequately portray a natural swimming movement and may have had unintended consequences on body position. Yanai (20) describes the interaction of the upper- and lower-body movements in freestyle swimming and how they contribute to "maintaining the bodyroll cycle" (p. 38). This suggests that the upper- and lower-body movements are coupled and necessary in swimming freestyle. When performing an arms-only condition, the legs would have been either left to drag behind the swimmer or needed to be supported (such as by a pull buoy). Letting them drag behind the swimmer would have resulted in increased drag because of the legs dropping further below the surface and creating a greater frontal area exposed to the oncoming flow of water. This would have increased the demand on the arms to overcome this drag and inflated the oxygen consumption costs. Conversely, if the legs were supported by a pull buoy, then the body position may have been artificially improved by raising the legs higher in the water than naturally occurring. This would reduce the frontal area exposed to the oncoming flow of water and reduce drag. In turn, this would artificially reduce oxygen consumption. Despite the attraction of using an arms-only condition, the inclusion of kicking was deemed to be necessary given the desire to study natural swimming movements. The choice to use whole body swimming in the current study precluded examination of separate upper- and lower-body contributions to swimming economy, but given the interaction of these components, it may not be possible to examine these contributions separately.
The relationship of cycle length and cadence to speed varies with mode of locomotion. For example, runners initially increase speed by increasing stride length while maintaining stride rate (7). Further increases in running speed are the result of increased stride rate while stride length is maintained. In contrast, Craig and Pendergast (3) found that swimmers sacrifice SL while increasing SR to increase speed. Thus, an emphasis in the coaching of competitive swimming has been to increase speed by increasing SL without sacrificing SR. Although the present study offers little insight as to adaptations associated with this type of training, it is important to understand the physiological cost and technique changes associated with such strategies. The current study suggests that, at least initially, acute increases in SL will result in substantially greater V˙O2, perhaps because of a substantially higher KR. Longitudinally, increases in swimming speed have been associated with increases in SL (2,18). This suggests that improvements in SL with experienced swimmers may develop naturally with training and improved fitness, thus limiting the need to expose the swimmers to the consequences of acute changes in SL.
Although the physiological responses to altering cadence did not follow the U-shaped pattern observed in other forms of locomotion, the absence of this pattern in the psychological response was not as clear. For the +20% SR condition, RPE increased 8% over that reported for the preferred SR. This change was characterized by a moderate effect size (ES = 0.41), suggesting an elevated perception of effort at the higher SR. Furthermore, when the RPE data were fit to a quadratic equation, the coefficient of the quadratic term was significantly different from zero (P = 0.003). This suggested that the U-shaped response was present in these data despite the lack of significant differences between the preferred SR condition and the higher SR conditions. Thus, at least in the analysis of RPE, it was possible that a type II error existed. If such errors occurred in the analyses, increasing the number of participants studied may have revealed psychological and possibly physiological responses at the higher SR that would better fit the U-shaped pattern seen in other forms of locomotion.
It was hypothesized that deviation from preferred SR in constant speed swimming would result in increased V˙O2 and that this deviation would be U-shaped. Our data provide limited support for these hypotheses. It appears that the selection of a cycle length and rate combination varies substantially between different forms of locomotion. In swimming as in cycling and arm-cranking exercise, the most economical cadence is not necessarily the preferred cadence. Therefore, as Martin et al. (11) point out, minimization of aerobic demand is not the sole organizing criterion for locomotion. For the present study, as indicated by the nonsignificant changes in oxygen uptake (and characterized by small ES) when SR was increased, swimmers preferred to swim freestyle at the lowest SR (or the longest SL) that did not require an increase in oxygen uptake.
The authors would like to thank Bob Wiskera, Dave Young, and Dr. Ben Levine for their assistance and support in the completion of this project. This project was funded through a Brown Faculty Fellowship from Southwestern University.
The results of the present study do not constitute endorsement by the American College of Sports Medicine.
1. Cavanagh PR, Williams KR. The effect of stride length variation on oxygen uptake during distance running. Med Sci Sports Exerc
2. Costill DL, Thomas R, Robergs A, et al. Adaptations to swimming training: influence of training volume. Med Sci Sports Exerc
3. Craig AB, Pendergast DR. Relationships of stroke rate, distance per stroke, and velocity in competitive swimming. Med Sci Sports Exerc
4. Högberg P. How do stride length and stride frequency influence the energy-output during running? Arbeitsphysiologie
5. Holt KG, Hamill J, Andres RO. Predicting the minimal energy costs of human walking. Med Sci Sports Exerc
6. Hreljac A, Martin PE. The relationship between smoothness and economy
during walking. Biol Cybern
7. Luhtanen P, Rahkila P, Rusko H, Viitasalo JT. Mechanical work and efficiency in ergometer bicycling at aerobic and anaerobic thresholds. Acta Physiol Scand
8. Marais G, Dupont L, Maillet M, Weissland T, Vanvelcenaher J, Pelayo P. Spontaneously chosen crank rate variations in submaximal arm exercise with inexperienced subjects: effects on cardiorespiratory and efficiency parameters. Int J Sports Med
9. Marais G, Weissland T, Robin H, Vanvelcenaher JM, Lavoie JM, Pelayo P. Physiological effects of variations in spontaneously chosen crank rate during sub-maximal and supra-maximal upper body exercises. Int J Sports Med
10. Marsh AP, Martin PE. The association between cycling experience and preferred and most economical cadences. Med Sci Sports Exerc
11. Martin PE, Sanderson DJ, Umberger BR. Factors affecting preferred rates of movement in cyclic activities. In: Zatsiorsky VM, editor. Biomechanics in Sport: Performance Enhancement and Injury Prevention (Volume IX of the Encyclopaedia of Sports Medicine)
. Oxford (UK): Blackwell Science Ltd.; 2000. p. 143-60.
12. Pawelczyk JA, McMinn S, Murchison B, Jacoby AC. Swimming economy
is optimized at a freely chosen distance per stroke. Med Sci Sports Exerc
13. Powers SK, Beadle RE Mangum M. Exercise efficiency during arm ergometry: effects of speed and work rate. J Appl Phys
14. Rosenthal R. Cumulating psychology: an appreciation of Donald T. Campbell. Psychol Sci
15. Swaine I, Reilly T. The freely-chosen swimming stroke rate in a maximal swim and on a biokinetic swim bench. Med Sci Sports Exerc
16. Toussaint HM, Meulemans A, de Groot G, Hollander AP, Schreurs AW, Vervoorn K. Respiratory valve for oxygen uptake measurements during swimming. Eur J Appl Physiol
17. Truijens MJ, Toussaint HM, Dow J, Levine BD. Effect of high-intensity hypoxic training on sea-level swimming performances. J Appl Physiol
18. Wakayoshi K, Yoshida T, Ikuta Y, Mutoh Y, Miyashita M. Adaptations to six months of aerobic swim training. Changes in velocity, stroke rate, stroke length and blood lactate. Int J Sports Med
19. Weissland T, Pelayo P, Vanvelcenaher J, Marais G, Lavoie JM, Robin H. Physiological effects of variations in spontaneously chosen crank rate during incremental upper-body exercise. Eur J Appl Physiol Occup Physiol
20. Yanai T. What causes the body to roll in front-crawl swimming? J Appl Biomater
21. Zarrugh MY, Radcliffe CW. Predicting metabolic cost of level walking. Eur J Appl Physiol