Journal of Applied Physiology

Effect of locomotor respiratory coupling on respiratory evaporative heat loss in the sheep

Pauline L. Entin, David Robertshaw, Richard E. Rawson


During galloping, many animals display 1:1 coupling of breaths and strides. Locomotor respiratory coupling (LRC) may limit respiratory evaporative heat loss (REHL) by constraining respiratory frequency (f). Five sheep were exercised twice each, according to a five-step protocol: 5 min at the walk, 5 min at the trot (trot1), 10 min at the gallop, 5 min at the trot (trot2), and 5 min at the walk. Rectal temperature (Tre), stride frequency, f, REHL, and arterial CO2 tension and pH were measured at each step. Tidal volume (Vt) was calculated. LRC was observed only during galloping. The coupling ratio remained at 1:1 while Vt increased continuously during galloping, causing REHL to increase from 2.9 ± 0.2 (SE) W/kg at the end of trot1 to a peak of 5.3 ± 0.3 W/kg. Tre rose from 39.0 ± 0.1°C preexercise to 40.2 ± 0.2°C at the end of galloping. At the gallop-trot2 transition, Vt fell and f rose, despite a continued rise in Tre. Arterial CO2 tension fell from 36.5 ± 1.1 Torr preexercise to 31.8 ± 1.4 Torr by the end of trot1 and then further to 21.5 ± 1.2 Torr by the end of galloping, resulting in alkalosis. In conclusion, LRC did not prevent increases in REHL in sheep because Vt increased. The increased Vt caused hypocapnia and presumably elevated the cost of breathing.

  • thermoregulation
  • entrainment
  • exercise

during galloping, a variety of species ranging from gerbils to rhinoceroses have been shown to exhibit a phase-locked, 1:1 coupling of breaths and strides (21). The apparent mechanical constraint on respiratory frequency (f) due to locomotor respiratory coupling (LRC) has led to speculation that both blood gases (4, 9) and the rate of respiratory evaporative heat loss (REHL) (18) during galloping might be the passive consequences of entrainment rather than the province of central regulation. Although it has been demonstrated that LRC is not the cause of hypercapnia and hypoxemia during maximal intensity exercise in the horse (2), the effect of LRC on thermoregulation remains unresolved. Locomotor respiratory coupling might compromise thermoregulation in those species that rely on panting for thermolysis because it may prevent f from being raised above the stride frequency (SF). Kamau (18) found that hyperthermia did not disrupt 1:1 coupling of f and SF during galloping in the dik-dik, despite the ability of that antelope to achieve a significantly higher f when panting at rest. Kamau presumed that REHL was limited by the entrained f; however, because REHL was not measured, the effect of LRC on thermoregulation remains speculative.

The purpose of the present study was to evaluate the effects of LRC on thermoregulation in a panting species, the sheep, by directly measuring REHL. Several distinct interactions between thermal drive and LRC are possible. If LRC is obligatory, then, as suggested by Kamau (18), REHL may be constrained by the relatively low f during LRC. Alternatively, an increase in tidal volume (Vt) would allow compliance with a 1:1 coupling ratio while meeting the need for increased REHL. An increase in Vt would, however, be expected to increase the work of breathing (7, 13) and to cause hypocapnia. If 1:1 LRC is not obligatory, then thermoregulatory respiratory patterns may prevail. In that case, thermal drive may disrupt entrainment such that breaths and strides become uncoupled and f is increased to augment REHL. Alternatively, entrainment may be maintained but at a higher coupling ratio, for example, 2:1 breaths per stride. Because thermoregulation has been demonstrated to take precedence over chemoregulation during hyperthermic exercise in the sheep (11), it was hypothesized that thermal drive would also override LRC in this species.

To distinguish among the competing hypotheses, f, SF, REHL, rectal temperature (Tre), and the arterial CO2 tension ( PaCO2 ) were measured in sheep exercising at three different gaits: the walk, trot, and gallop. LRC was not consistently observed in walking and trotting dogs (1) or horses (15) and was not expected during these gaits in the sheep. At the gallop, horses (8, 19) and dogs (21) consistently exhibited LRC, and hence LRC was expected, at least initially, during galloping in the sheep. The sheep were walked and trotted both before and after galloping so that the effect of heat load, which rises rapidly during high-intensity exercise, could be differentiated from the effect of gait per se.



Five Dorset ewes [55 ± 6 (SD) kg] were trained to walk, trot, and gallop on a motorized treadmill. The walk was defined as a gait in which the limbs moved in lateral pairs (e.g., left foreleg, left hind leg) and at least three feet were on the ground at all times. The trot was defined as a gait in which the limbs moved in diagonal pairs (e.g., left foreleg, right hind leg). A suspensory phase within each stride, during which all four hooves were off the ground, was the defining characteristic of the gallop. These gaits could be easily differentiated by a practiced observer. The sheep were housed individually in 1.8-m2 pens in a room with a 14:10-h light-dark cycle. All sheep were fed daily ∼800 g of hay and 200 g of grain (Early Market Lamb Pellets, Agway, Syracuse, NY). Water was provided ad libitum. A minimum of 16 h separated feeding and the start of an experimental trial.

Surgical preparation.

Several months before the experiments, each sheep was prepared with an externalized carotid artery by using the method described by Hales and Webster (14). Briefly, the sheep was placed under halothane anesthesia, and a semicircular incision was made on the right lateral aspect of the neck. Approximately 10 cm of the carotid artery were isolated from the vagosympathetic nerve, and any branches of the artery in that region were ligated. An oval plastic plate (4 cm long and 0.5 cm thick) was sutured in place under the artery. The carotid artery then rested superficially, in a groove extending the length of the plate. On the morning of an experiment, with the sheep under lidocaine local anesthesia, the carotid artery was acutely cannulated with an 18-gauge 3.8-cm plastic cannula (IV Cath, Becton-Dickenson, Rutherford, NJ). The cannula was removed at the end of the experimental trial.

Experimental protocol.

Experiments were conducted in accordance with the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The research protocol was approved by the Institutional Animal Care and Use Committee of Cornell University.

Three exercise intensities were selected for the purpose of eliciting three locomotor gaits: walking, trotting, and galloping. These three gaits were incorporated into five sequential phases for a total exercise duration of 30 min. In chronological order, the phases were as follows: 5 min at the walk (walk1), 5 min at the trot (trot1), 10 min at the gallop (gallop), 5 min at the trot (trot2), and 5 min at the walk (walk2) (Table 1). The treadmill incline was held at 0° throughout. Within each individual exercise trial, the treadmill speeds during trot2 and walk2 were matched (±0.1 m/s) to the speeds during trot1 and walk1, respectively. During galloping, the treadmill speed was increased from that of trot1 until the sheep switched gaits from the trot to the gallop. The speed was then maintained at that level for the 10-min galloping phase. The mean treadmill speeds during each phase are displayed in Table 1.

View this table:
Table 1.

Duration, speed, stride frequency, respiratory rate, breaths per stride ratio, and stride frequency-respiratory rate correlation of each of the 5 phases of exercise

To ensure that the sheep would be able to complete the 10-min galloping phase, the ewes were put on a physical training regimen for at least 2 wk before testing. Training consisted of 30-min exercise bouts of walking and/or trotting, usually three times per week. After training, each sheep completed the incremental exercise test twice. A minimum of 14 days separated the two trials completed by one individual sheep. The mean air temperature during the experimental trials was 22.4 ± 0.8 (SD) °C (n = 10). During all trials and training runs, a window fan was placed behind the sheep from preexercise through the postexercise period. The fan provided a fixed airflow that was not matched to treadmill speed.

An auxiliary experiment was performed to determine the relationship between heat production and REHL during uncoupled respiration. Two sheep were used. The treadmill speed and incline were manipulated so as to vary heat production from 5 to 16 W/kg while the only gaits elicited were the walk and trot. REHL was measured after 15 min at each work level. Thirteen measurements were used to construct a regression equation for the relationship between heat production and REHL after 15 min of unentrained exercise.

Data collection.

The sheep wore a lightweight plastic face mask through which air was drawn at a rate of 730 l/min by a vacuum cleaner located in an adjacent room (Fig. 1). Oxygen consumption (V˙o 2) was measured continuously via an open-flow system that has been described previously (17). The open flow system has no valves and no dead space. Heat production was calculated fromV˙o 2 by using a caloric equivalent of 20.1 J/ml O2. A copper-constantan thermocouple encased in plastic was used to measure Tre. Samples of inspired and expired air were pumped continuously to two temperature and humidity sensors (model HMI-32, Vaisala Woburn, MA). The humidity sensors were calibrated according to the known vapor pressures above saturated solutions of sodium chloride and lithium chloride. The temperature and humidity data, along with Tre, were collected by a data-acquisition system and stored on a computer at 1 Hz. The water content of inspired and expired air was computed directly from the temperature and humidity measurementsWater content enteringor leaving the mask(g/min)=(SatVP)(rh/100)(18)(730)R(T) Equation 1 where Sat VP is the saturation vapor pressure of water at the relevant temperature, rh is the relative humidity, 18 is the molecular weight of water, 730 is the flow rate of air through the system, R is the universal gas constant, and T is the temperature of the air entering or leaving the mask.

Fig. 1.

Experimental setup. Sheep wore a plastic face mask through which room air was drawn at a rate of 730 l/min. Temperature and humidity sensors sampled room air and expired air. A sample of room air was routed to CO2 and O2 sensors. Arterial blood samples were drawn from the carotid artery. Rectal temperature was measured by thermocouple. Diff Press Trans, differential pressure transducer.

Respiratory evaporative water loss (REWL) was calculated by subtracting the water content of inspired air from that of expired air. REWL was converted to REHL by assuming a specific heat of 1.0 J ⋅ g−1 ⋅ °C−1and a latent heat of vaporization of 2,447 J/g. The accuracy of the system was checked by evaporating a known mass of water into the mask. This procedure revealed a 1–9% underestimation of the actual water loss.

Minute ventilation (V˙e) also was derived from REWLV˙E=(REWL)(R)(Texpair)(SatVP)(18) Equation 2 where Texpair is the expired air temperature and Sat VP is the saturation vapor pressure at the expired air temperature.

Expired air temperature was measured in separate experiments on six sheep during exercise at 2 m/s (0° incline) by using a 32-standard wire gauge copper-constantan thermocouple placed just outside of one nostril. The expired air temperature averaged 32.0 ± 0.9 (SE) °C. In sheep, expired air is known to be saturated with water vapor (16).

The value of f was calculated from continuous measurements of esophageal pressure. A finger from a latex glove was tied around one end of Tygon plastic tubing (1.78 mm OD) to function as an inflatable balloon. The other end of the tubing was connected to a differential pressure transducer. The balloon was inserted ∼60 cm into the esophagus via the nose and then was inflated with 1 ml of air. It was assumed that the balloon rested at a position approximately level with the heart. Output from the pressure transducer was amplified and displayed on a computer. Respiratory rate was computed from the continuous record of inspiration and expiration. Vt was calculated by dividingV˙e by f. SF was determined visually.

Arterial blood samples were collected twice during the preexercise period: ∼6 and 1 min before the start of exercise. During exercise, blood samples were collected at 5, 7.5, 10, 13, 16, 19, 22.5, 25, and 30 min. These times corresponded to the end of walk1 (5 min), the middle and end of trot1 (7.5 and 10 min, respectively), the gallop (13, 16, and 19 min), the middle and end of trot2 (22.5 and 25 min, respectively), and the end of walk2 (30 min). Blood samples were also collected 5, 10, and 20 min postexercise. The samples were capped, placed on ice, and later analyzed for PaCO2 and pH on a blood-gas analyzer (Radiometer ABL-30, Copenhagen, Denmark). The analysis temperature was set at the Tre of the sheep at the time the sample was taken. The accuracy of the blood-gas analyzer was verified by using tonometered sheep or horse blood.


The mean f of each phase of the exercise bout was calculated by averaging all of the measurements taken during that phase, usually 10 measurements per trial for walk1, trot1, walk2, and trot2 and 17–19 measurements for the galloping phase. The ratio of f to SF (f/SF) for each phase of exercise was computed by dividing the mean f by the SF of that phase. For each step of the exercise protocol, the strength of the correlation between f and SF was tested by regressing the f of each trial on the SF of that trial (10 points, 2 per sheep).

The V˙o 2 during each phase of exercise, as well as pre- and postexercise, was taken as the mean of at least 1 min of measurement during which time theV˙o 2 was constant. (Transition periods between the phases and at the onset and termination of exercise were excluded.) Representative values of Tre and REHL were obtained by averaging over the last 1 min of each phase of exercise and during the last 1 min preexercise. The postexercise Tre and REHL values represent the mean from 9 to 10 min postexercise.

For all variables, the values from the two trials on each sheep were averaged first, and then those averages were used to calculate the mean value over the five sheep. Results are reported as means ± SE, unless otherwise noted. Comparisons among time points or phases of exercise were made within sheep by using a pairedt-test. Statistical significance was recognized at P < 0.05.


The mean V˙o 2 preexercise was 6.1 ± 1.0 ml ⋅ kg−1 ⋅ min−1(n = 4;V˙o 2 was not measured in 1 sheep due to equipment failure).V˙o 2 rose to 14.7 ± 1.0 ml ⋅ kg−1 ⋅ min−1during walk1, 23.9 ± 1.0 ml ⋅ kg−1 ⋅ min−1during trot1, and 39.4 ± 1.9 ml ⋅ kg−1 ⋅ min−1during the gallop. TheV˙o 2 returned to 24.6 ± 0.5 ml ⋅ kg−1 ⋅ min−1during trot2 and to 16.1 ± 1.0 ml ⋅ kg−1 ⋅ min−1during walk2. At ∼10 min postexercise, the meanV˙o 2 was 5.5 ± 0.2 ml ⋅ kg−1 ⋅ min−1. Heat production was calculated from the V˙o 2 values (Fig.2).

Fig. 2.

Heat production and respiratory evaporative heat loss (REHL) preexercise; during the walk, trot, and gallop; and 10 min postexercise. Values are means ± SE of 4 sheep for heat production and of 5 sheep for REHL.

In all sheep, REHL increased with each increase in treadmill speed from walk1 through the gallop (Fig. 2). From a peak value of 5.3 ± 0.03 W/kg at the end of galloping, REHL decreased by an average of 0.26 ± 0.07 W/kg within 1 min of trot2. All values of REHL during trot2 and walk2 were greater than those during trot1 and walk1. Similarly, the mean REHL postexercise was nearly twice that during preexercise, and it exceeded the postexercise heat production (Fig. 2).

Tre rose throughout exercise from 39.0 ± 0.1°C preexercise to 40.6 ± 0.2°C at the end of walk2 (Fig. 3). the most rapid rate of change in Tre was observed during galloping when Tre increased at an average rate of 0.1°C/min. Over the first 10 min of recovery from exercise, the mean Tre decreased from 40.6 ± 0.2 to 40.3 ± 0.2°C.

Fig. 3.

Rectal temperature preexercise; during the walk, trot, and gallop; and 10 min postexercise. Values are means ± SE of 5 sheep.

SF increased with increasing treadmill speed (Table 1). There was no significant difference in the SF between walk1 and walk2 or between trot1 and trot2.

The value of f varied considerably both within and between sheep during all phases of exercise but particularly during walking and trotting (Fig. 4). In four of five sheep, f increased over the course of walk1 and trot1. In all trials on all sheep, f neither increased nor decreased over the course of galloping, although random variation from minute to minute was observed (Fig.5). Some of the random variation in f was attributable to swallows (Fig. 6). During trot2, f either rose continuously or rose initially and plateaued at a value higher than that maintained during galloping. In 9 of 10 trials, f was higher during walk2 than during trot2. In all cases, f was higher during walk2 and trot2 than during walk1 and trot1 (Table 1, Fig. 4). Often f rose initially at the termination of exercise to a new peak and then decreased gradually (Fig. 5).

Fig. 4.

Respiratory rate plotted by stride frequency during 5 min at walk (walk1), 5 min at trot (trot1), 10 min at gallop (gallop), 5 min at trot (trot2), and 5 min at walk (walk2). Line of unity corresponds to a breaths to strides ratio of 1:1. Each data point represents a single exercise trial of an individual sheep.

Fig. 5.

Tidal volume, respiratory rate, and REHL preexercise; during the walk, trot, and gallop; and postexercise in 1 sheep. Note that respiratory rate remained constant throughout the period of galloping while tidal volume and REHL rose. At transition from galloping to trot2, tidal volume fell and respiratory rate rose, resulting in a 0.3 W/kg decline in REHL.

Fig. 6.

Esophageal pressure tracing of 1 sheep. Transition from galloping to trot2 is shown. Note that respiratory rate is higher during trot2 than during gallop. Large positive deflection immediately after gait transition represents a swallow.

The mean f/SF (breaths to strides) during each phase of exercise is shown in Table 1. During walk1, trot1, walk2, and trot2, there was no significant correlation between f and SF (Table 1). In contrast, during galloping, SF explained 73% of the variance in f (P < 0.004; Table 1). The relationship between f and SF during each phase of exercise can be seen in Fig. 4. Although the mean f/SF ratio was near 1 during walk1 and trot1, LRC probably did not occur because, individually, only one sheep exhibited a ratio near 1 for these two phases of exercise (Fig. 4). Moreover, f increased throughout walk1 and trot1 in both trials on this sheep, suggesting that the f/SF ratio near unity was a chance occurrence. In the other sheep, the ratio between f and SF during walk1 and trot1 ranged from 0.3 to 1.5 (Fig. 4). In contrast, at the gallop, the f/SF ratio of all sheep during all trials was between 0.93 and 1.06 (Fig. 4).

Vt was variable during preexercise and during walk1 and trot1. In all sheep, Vt increased continuously during galloping (Fig. 5). The peak value of Vt attained during galloping varied between sheep, ranging from 1.1 to 1.8 liters. At the onset of trot2, Vt fell immediately from the maximum level during galloping. At 10 min postexercise, Vt ranged between 450 and 650 ml.

Preexercise, the mean PaCO2 was 36.6 ± 1.1 Torr (Fig. 7). The PaCO2 fell significantly to 31.8 ± 1.3 Torr by the end of trot1. During the gallop, the PaCO2 fell further to 21.5 ± 1.2 Torr. The mean PaCO2 did not change significantly from that value through trot2 and walk2 (Fig. 7). Within 5 min of recovery from exercise, PaCO2 increased significantly to a mean value of 29.8 ± 2.1 Torr. By 20 min postexercise, the PaCO2 had risen an additional 4 Torr to 34.2 ± 1.4 Torr (Fig. 7).

Fig. 7.

Arterial carbon dioxide tension ( Embedded Image ) preexercise; during the walk, trot, and gallop; and 5, 10, and 20 min postexercise. Values are means ± SE of 5 sheep.

Arterial pH rose from 7.45 ± 0.02 preexercise to 7.50 ± 0.03 at the end of trot1 (Fig.8). During galloping, one sheep exhibited a metabolic acidosis as evidenced by a pH below that of preexercise. The other four sheep were alkalotic during galloping. In all sheep, arterial pH remained above the preexercise value during trot2 and walk2 (Fig. 8). Arterial pH fell back to 7.44 ± 0.02 at 10 min postexercise (Fig. 8).

Fig. 8.

Arterial pH preexercise; during the walk, trot, and gallop; and 5, 10, and 20 min postexercise. Values are means ± SE of 5 sheep.

A linear relationship was found between heat production and REHL after 15 min of unentrained exercise (Eq.3 )REHL(W/kg)=0.35[heat production(W/kg)] +0.50;R2=0.93 Equation 3The REHL predicted at the level of heat production observed during galloping, 13.2 W/kg, was 5.1 W/kg.


A 1:1 ratio of f to SF was observed during galloping, but not during walking or trotting. Although temporal (“phase-locked”) entrainment between the respiratory cycle and the locomotor cycle (for example, between expiration and forelimb ground contact) could not be verified in the absence of a continuous record of limb movements, LRC was indicated by 1) an f/SF ratio near 1 (0.93–1.06) during galloping in all experimental trials (Fig. 4) and 2) the lack of increase or decrease in f during galloping (Fig. 5). The interruption of breathing by swallowing (Fig. 6) can account for the failure of f and SF to reach parity during galloping in some sheep. The observation of LRC during galloping, but not during walking or trotting, is consistent with similar observations in dogs (1) and horses (8, 15).

LRC did not constrain REHL. The peak level of REHL reached during galloping was not significantly different from that expected after 15 min of uncoupled respiration at the same level of heat production (i.e., V˙o 2), on the basis of the correlation between heat production and REHL established in two sheep under the same laboratory conditions (Eq.3 ). During uncoupled respiration, sheep usually increase f to escalate REHL; however, during entrainment, Vt was increased. Although part of the increase in Vt may be accounted for by the greater metabolic cost of galloping vs. trotting, two lines of reasoning support the proposition that the increase in Vt exceeded that required to raise V˙o 2. First, the progressive decline in PaCO2 during galloping indicates that alveolar ventilation exceeded CO2 production (i.e., the animals hyperventilated). Second, REHL andV˙o 2 changed with different time courses at the onset and offset of galloping. Within 2 min of the change in exercise intensity from trot1 to the gallop, a new steady-state V˙o 2 was reached, yet REHL, and by inference Vt, continued to increase throughout the entire 10-min duration of galloping. The abrupt 0.1–0.5 W/kg fall in REHL from the end of galloping to the beginning of trot2 (Fig. 5) probably represents the component of REHL that can be attributed to the greater gas-exchange requirements of galloping vs. trotting, because metabolic rate, but not the cumulative heat storage, fell during that transition. If this assumption is correct, then <25% of the total increase in REHL during galloping can be attributed to the increase in metabolic rate. Alveolar ventilation was not stimulated by metabolic acidosis, because the arterial pH remained above the preexercise level throughout galloping (Fig. 8). Consequently, the majority of the increase in Vt during galloping must be attributed to the thermal drive to pulmonary ventilation. The increase in Vt during LRC is reminiscent of the increase in Vt seen during “second-phase breathing” in severely heat-stressed sheep at rest (14) and during exercise (10). Similarly, Bayly et al. (5) concluded that thermoregulatory drive stimulated the progressive increase in Vt, and resulting fall in PaCO2 , that they observed in exercising horses.

Whether LRC is physically obligatory during galloping cannot be determined from the results of the present study. The observation that thermal drive did not disrupt LRC in the sheep is consistent with the observations of Kamau (18) on the galloping dik-dik but contrasts with the results of Gillespie et al. (12), who found that 1:1 (f/SF) coupling could be interrupted by increased inspired CO2 in one galloping horse. Whether or not thermal drive can disrupt LRC during galloping in panting species remains to be determined.

Even if LRC is not obligatory, Bramble and Carrier (6) suggested that LRC is energetically favored by the contribution of locomotion to the generation of respiratory airflows. In the guinea fowl, the f at the preferred running SF has been shown to be equal to the natural resonance frequency of the respiratory system (20). In these animals, uncoupled respiration might be more costly. However, the energetic advantage of LRC in the running or galloping mammal has yet to be established (9). The contribution of limb movements to respiratory airflow has been found to be negligible in trotting dogs (1) and in walking and running humans (3). The immediate reversion to an uncoupled, high-f respiratory pattern by the sheep at the transition from the gallop to trot2 (Fig. 5) suggests that a high-f, moderate-Vt pattern has less impact on pH balance or is less energetically expensive than a moderate-f, high-Vt ventilatory pattern. The energetic cost of the increased Vt during LRC in the hyperthermic sheep, as in the exercising horse (5, 7) and heat-stressed ox (13), may have negated any energetic advantage of the coupling per se.

The defense of body temperature competes with both the minimization of ventilatory work and pH balance in the exercising sheep. During galloping, augmentation of REHL was achieved without disruption of LRC but at the expense of alkalosis and presumably a higher work of breathing (5, 7, 13). Neither a steady-state REHL nor Vt was observed during galloping; however, presumably a maximum Vt exists. A higher coupling ratio, such as 2:1 (f/SF), would provide an alternate means to increase REHL; however, despite the presumably high cost of breathing associated with a large Vt, only 1:1 coupling was observed.


The authors thank Katherine Streeter, Victoria Wildman, and Darren Wilson for their assistance with this research.


  • Address for reprint requests and other correspondence: P. L. Entin, Dept. of Medicine, Box 0623A, 9500 Gilman Dr., Univ. of California, San Diego, La Jolla, CA 92093-0623 (E-mail:pentine{at}

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.


View Abstract