INTRODUCTION

Top Abstract Introduction Methods Results Discussion References

WHEN THE INSPIRATORY muscles contract against a high load, they are able to maintain the target level of contraction for a limited period of time, i.e., until "task failure," after which the load cannot be sustained any longer. Task failure is believed to be due to various degrees of central inhibition (termed central fatigue) (1), contractile fatigue (1), and occasionally lack of motivation. The contractile fatigue that occurs with a high load is believed to be due to mechanical damage to the muscle (26, 36). The threshold of loading necessary to produce such damage has not been defined precisely. It seems reasonable to expect that the length of time that a load is sustained, i.e., endurance time, may also be an important factor in the development of muscle injury and the functional consequences.

The precise contribution of progressively developing contractile fatigue to task failure is unclear. In seven subjects performing isocapnic hyperpnea, Babcock et al. (3) reported significant diaphragmatic fatigue [~26% decrease in transdiaphragmatic twitch pressure (Pdi_tw)] without evidence of task failure. Their study does not elucidate completely the relationship between the progressive development of contractile fatigue and the occurrence of task failure, however, because inspiratory loading was not maintained to the point of task failure and the study was not designed to investigate the progression of diaphragmatic fatigue during loading.

Bellemare and Bigland-Ritchie (5) reported that inspiratory resistive loading produced a decrease in Pdi_tw during the first 40% of endurance time, but thereafter Pdi_tw showed no further decrease. This result is surprising and raises the possibility that contractile fatigue is not a progressive process, despite continued diaphragmatic loading. This interpretation is not tenable because of two limitations in the study design. 1) Forceful diaphragmatic contractions during inspiratory resistive loading induce fatigue and twitch potentiation (14). Because of these opposing effects, it is difficult to know whether the failure of Pdi_tw to decrease progressively during loading indicated that contractile fatigue is truly a nonprogressive process or is due to the offsetting effect of twitch potentiation (8). 2) Inspiratory loading results in a decrease in lung volume (9), and the associated increase in diaphragmatic length will cause Pdi_tw to increase (15), further confounding data interpretation. Recently, Eastwood et al. (9) evaluated the influence of progressive threshold loading on diaphragmatic contractility, assessed by Pdi_tw corrected for load-induced changes in end-expiratory lung volume (EELV). Unlike Bellemare and Bigland-Ritchie (5), they noted that Pdi_tw increased progressively until just before task failure, at which point Pdi_tw decreased. That is, impaired contractility and task failure appeared to occur almost simultaneously (9). However, as in the case of the investigation of Bellemare and Bigland-Ritchie (5), the influence of twitch potentiation on Pdi_tw was not assessed.

Another unanswered question is whether the pattern of respiratory muscle recruitment in the presence of increased ventilatory demands is affected by the progression (or lack of progression) of contractile fatigue to task failure. After the induction of diaphragmatic fatigue, Yan et al. (35) noted that the respiratory system responded to increased ventilatory demands (CO₂ rebreathing) by recruiting rib cage muscles. This interpretation is confounded by the fact that the response to CO₂ was tested shortly after the completion of the inspiratory resistive loading run (i.e., 5 min); at this time, high-frequency fatigue, which could modulate the recruitment of respiratory muscles during increased ventilatory demands, is expected to be present (2). In other words, by performing CO₂ rebreathing shortly after task failure, it is not possible to discriminate between the effects of high- and low-frequency fatigue or their combination on the diaphragm. As the same investigators pointed out (34), high-frequency fatigue may have more limited physiological and clinical relevance than low-frequency fatigue. Second, their study was not designed to elucidate the adaptive response of the respiratory muscles to increased ventilatory demands during progressive inspiratory resistive loading.

The purpose of our study was threefold: 1) to determine whether inspiratory resistive loading sustained for periods of increasing duration produces a parallel decrease in diaphragmatic contractility, ever before task failure, 2) to determine whether the pattern of respiratory muscle recruitment during CO₂ rebreathing is altered as the period of inspiratory resistive loading is progressively increased, and 3) to determine whether a 20-h period of recovery after task failure could normalize fatigue-induced changes in the pattern of respiratory muscle recruitment during increased ventilatory demands.

	METHODS

Top Abstract Introduction Methods Results Discussion References

Subjects. Ten nonsmoking healthy men (26-36 yr of age, mean 30 yr) volunteered for the study, which was approved by the Human Studies Subcommittee of Edwards Hines, Jr., Veterans Administration Hospital. Informed consent was obtained from all subjects.

Flow, volume, and pressure measurements. Inspiratory flow () was measured with a heated Fleisch pneumotachograph (Hans Rudolph, Kansas City, MO) connected to a differential pressure transducer (model MP-45, Validyne, Northridge, CA). Volumes were obtained by electronic integration of the signal. Inspiratory and expiratory timing were based on the signal. Esophageal (Pes) and gastric (Pga) pressures were separately measured with two thin-walled, latex balloon-tipped catheters (Erich Jaeger, Wurzberg, Germany) coupled to pressure transducers (model MP-45, Validyne). An esophageal balloon containing 1 ml of air was positioned in the midesophagus using the occlusion technique (4); a gastric balloon containing 2 ml of air was advanced 70 cm from the nares. Transdiaphragmatic pressure (Pdi) was obtained by electronic subtraction of Pes from Pga. Airway pressure was measured at the mouthpiece using a tap connected to a third transducer (model MP-45, Validyne). Maximal Pdi (Pdi_max) was recorded while the subjects performed a maximal Müller expulsive effort of 1-s duration against an occluded airway at EELV (16). Oscilloscope recordings of Pdi provided visual feedback.

Electromyography. Compound diaphragmatic motor action potentials (CDAPs) were recorded bilaterally with surface electromyogram (EMG) electrodes placed at the seventh and eighth intercostal spaces in the anterior axillary line. The subject's skin was marked with indelible ink to ensure that the electrodes were placed at the same location during the ~20-h follow-up study performed in 6 of the 10 subjects. All EMG signals were amplified, band-pass filtered (bandwidth 10 Hz-1 kHz; Gould, Valley View, OH), and displayed on a storage oscilloscope (Gould, Ilford, UK).

CO₂ rebreathing. To determine whether the pattern of respiratory muscle recruitment was altered when ventilatory demands were increased, the subjects rebreathed CO₂ according to the method of Read (25). Subjects breathed through a mouthpiece connected to a 7-liter bag filled with a balance of 7% CO₂ and O₂. The rebreathing bag remained flaccid so that the pressure within it was atmospheric. End-tidal PCO₂ (PET_CO2) was measured at the mouthpiece (CO₂SMO, Novametrix, Wallingford, CT). The procedure was terminated when PET_CO2 reached ~70 Torr, which generally took 3-4 min. During the whole protocol, O₂ saturation was measured with a pulse oximeter (CO₂SMO, Novametrix). During CO₂ rebreathing, as during the rest of the experiment, all subjects sat upright in an armchair with a high back. An investigator always remained in front of the subject to ensure that the subject sat straight, with his back against the back of the chair, while resting the arms on his lap.

Phrenic nerve stimulation. Magnetic stimulation of both phrenic nerves was performed using a magnetic stimulator (model 200, Magstim) with a 90-mm coil (P/N 9784-00). This device stimulates neuromuscular structures by inducing electrical currents in the tissue secondary to a time-varying magnetic field of brief duration (<1-ms total pulse duration) (29). At maximal output of the stimulator, the magnetic field is 2.0 T. To achieve stimulation of the phrenic nerves, the subject's neck was flexed and the coil was placed over the cervical spine. While the subject relaxed at EELV, the site of optimal stimulation was determined by moving the coil between C₅ and C₇ (29). Stimulus maximality was then assessed by progressively increasing the intensity of the stimulus until Pdi_tw and CDAPs displayed no further increases (29). This position was marked with indelible ink, and all subsequent stimulations were performed at this point. Subjects were studied without abdominal binding (14), waist belts were removed, trousers were unbuttoned, and alterations in abdominal compliance during stimulations were minimized by instructing the subjects to relax their diaphragm and abdominal wall muscles during each stimulation. Relaxation was confirmed by the absence of diaphragmatic EMG activity. Respiratory inductive plethysmography (Non-Invasive Monitoring Systems, Miami Beach, FL) was used to identify EELV.

Air hunger and effort scores. The intensity of dyspnea, defined as "the unpleasant sensation of labored or difficult breathing" (24), was evaluated using a modified Borg scale every minute during inspiratory resistive loading until task failure. Subjects were asked to score their sensation of air hunger and effort, respectively, by answering the following questions (30): 1) How much air hunger do you feel? 2) How much effort does your breathing require?

Inspiratory resistive load. Subjects were instructed to sustain an inspiratory resistive load that consisted of breathing through a mouthpiece attached to a one-way valve (Hans Rudolf), the inspiratory side of which was connected to a variable alinear resistor. The subject was instructed to breathe through the load while maintaining a Pdi of 60% of Pdi_max. Resistance was achieved by a screw clamp on a compressible tube, adjusted to help the subject reach the predetermined fraction of Pdi_max; the expiratory side, where PET_CO2 was monitored, was not loaded. The subject was instructed to maintain a constant Pdi throughout inspiration, as reflected by a square-wave pattern on the oscilloscope. The duty cycle was 0.50, and the frequency (f) was 15 breaths/min. Each subject was allowed to choose his own tidal volume (VT), and the relative contributions of Pga and Pes to Pdi were not controlled (14, 35). Subjects were encouraged to maintain the target Pdi in the desired fashion and to continue with the protocol.

Experimental protocol. After placement of all transducers, Pdi_max (best of 5 efforts) was determined. The phrenic nerves were stimulated twice at relaxed EELV immediately after each Pdi_max maneuver, and thus the recorded twitches represent potentiated Pdi_tw (14, 19, 33) (Fig. 1). To obtain nonpotentiated values of Pdi_tw, the subjects breathed quietly for 20 min, and then 10 twitches were recorded at EELV (14, 19, 33). At 3-5 min after the last nonpotentiated Pdi_tw, the ventilatory demands were increased by CO₂ rebreathing.

View larger version (22K): [in this window] [in a new window]   Fig. 1.   Schematic representation of experimental protocol. Respiratory muscle testing before loading (extreme left) and ~20 h after last period of loading (extreme right) consisted of maximal transdiaphragmatic pressure (Pdimax) followed by potentiated transdiaphragmatic twitch pressure (Pditw) and, after a 20-min rest, by nonpotentiated Pditw and CO2 rebreathing. Respiratory (Resp) muscle testing performed 20 min after each period of loading also included recordings of Pdimax and potentiated Pditw.

Data analysis. All signals were recorded and digitized at 2,000 Hz using a 12-bit analog-to-digital converter (CODAS, DATAQ Instruments, Akron, OH) connected to a computer (EMPAC, Fremont, CA). Individual twitch responses were rejected from analysis according to previously described criteria (14).

	RESULTS

Top Abstract Introduction Methods Results Discussion References

Air hunger and effort scores. Air hunger increased after the onset of each inspiratory resistive loading protocol (Fig. 2). During the first 2 min of loading, the increase in air hunger was the same for the three protocols; during the 2nd-4th min of loading the rise in air hunger was the same during the 4-min protocol and task-failure protocol. At task failure, the air hunger score was 7 ± 1 arbitrary units. The sense of effort also increased after the onset of each protocol. During the task-failure protocol, the sense of effort was greater at the 4th min than at the 4th min of the 4-min protocol (P < 0.05; Fig. 2). In all instances, the sense of effort was maximal at task failure.

View larger version (13K): [in this window] [in a new window]   Fig. 2.   Scores for air hunger and effort recorded every minute during periods of inspiratory resistive loading (IRL) lasting 2 min, 4 min, and until task failure. A: sense of air hunger increased after onset of each loading protocol; rate of rise did not change as period of loading was progressively increased. B: sense of effort increased after onset of each loading protocol; effort was greater at 4th min of task-failure protocol than at 4th min of 4-min protocol. Values are means ± SE; AU, arbitrary units. * Significantly different from baseline, P < 0.05.

Pdi_max. Baseline Pdi_max was 194 ± 7 (SE) cmH₂O (range 165-230 cmH₂O). After 2 min of inspiratory resistive loading, Pdi_max decreased to 161 ± 5 cmH₂O (P < 0.01; Fig. 3). After 20 min of resting breathing, Pdi_max increased to 176 ± 7 cmH₂O (P < 0.05), which was still less than baseline (P < 0.01). After the 4-min period of loading, Pdi_max decreased to 158 ± 6 cmH₂O, which was less than that recorded at baseline (P < 0.01) and just before this second period of loading (P < 0.01). After 20 min of resting breathing, Pdi_max increased to 177 ± 6 cmH₂O (P < 0.01), which was less than baseline but not different from that recorded just before this second period of loading. At the conclusion of the task-failure protocol, Pdi_max decreased to 153 ± 8 cmH₂O, which was less than that recorded at baseline (P < 0.01) or just before the task-failure protocol (P < 0.01). After 20 min of resting breathing, Pdi_max increased to 171 ± 6 cmH₂O (P < 0.01), which was less than baseline but not different from that recorded just before the task-failure protocol. After ~20 h, Pdi_max was 172 ± 16 cmH₂O (n = 6), which was less than the value at baseline (P < 0.05; Fig. 3).

View larger version (22K): [in this window] [in a new window]   Fig. 3.   Pdimax recorded before and after inspiratory resistive loading (hatched bars) sustained for 2 min, 4 min, and until task failure. After each period of loading, Pdimax was recorded immediately and after 20 min (n = 10) and at ~20 h after conclusion of task-failure protocol (n = 6). After only 2 min of loading, Pdimax was reduced from baseline. Nadir of Pdimax was reached just after conclusion of loading to task failure. Pdimax values at ~20 h remained lower than baseline. Values are means ± SE. Significantly different from baseline: * P < 0.05; ** P < 0.01.

Nonpotentiated Pdi_tw. Nonpotentiated Pdi_tw at baseline was 35.4 ± 2.4 cmH₂O (range 22.1-44.4 cmH₂O). Inspiratory resistive loading for 2 min did not produce any significant change in nonpotentiated Pdi_tw (Fig. 4). After the 4-min period of loading, the nonpotentiated Pdi_tw was 32.0 ± 2.2 cmH₂O, which was less than that at baseline (P < 0.01). Loading to task failure decreased nonpotentiated Pdi_tw to 30.0 ± 1.7 cmH₂O at 20 min, which was less than the values at baseline (P < 0.01) and 20 min after the conclusion of 2 min of loading (P < 0.01; Fig. 4). After ~20 h, nonpotentiated Pdi_tw was 30.1 ± 1.8 cmH₂O (n = 6), which was less than the value at baseline (P < 0.01) and unchanged from the Pdi_tw value recorded ~20 h earlier (Fig. 4). The amplitude of the right and left CDAPs was the same throughout this ~20-h period (Fig. 5), indicating a constant level of diaphragmatic recruitment.

View larger version (27K): [in this window] [in a new window]   Fig. 4.   Pattern of decrease and recovery of potentiated and nonpotentiated Pditw after inspiratory resistive loading (hatched bars) sustained for 2 min, 4 min, and until task failure. After each period of loading, potentiated Pditw was recorded immediately and after 20 min (n = 10) and at ~20 h after conclusion of task-failure protocol (n = 6). Potentiated Pditw was reduced from baseline values after first 2 min of loading, and nadir was reached 20 min after conclusion of inspiratory resistive loading to task failure. Potentiated Pditw at ~20 h from task failure remained lower than baseline. Nonpotentiated Pditw was reduced after 4-min period of loading, and at ~20 h from task failure it remained lower than baseline. Values are means ± SE. ** Significantly different from baseline, P < 0.01.

View larger version (26K): [in this window] [in a new window]   Fig. 5.   Changes in potentiated (A) and nonpotentiated (B) Pditw and associated right and left compound diaphragmatic action potentials (CDAPs) after inspiratory resistive loading (hatched bars) to task failure (n = 6). Twitches were recorded before loading (baseline) and at 20 min and ~20 h after task failure. Potentiated and nonpotentiated Pditw were reduced from baseline after inspiratory resistive loading, and at ~20 h from task failure they remained lower than baseline. Degree of diaphragmatic recruitment elicited during potentiated and nonpotentiated Pditw recordings remained constant over time, as demonstrated by constancy in CDAP amplitudes. Values are means ± SE. ** Significantly different from baseline, P < 0.01.

Potentiated Pdi_tw. Potentiated Pdi_tw at baseline was 54.1 ± 2.7 cmH₂O (range 34.4-64.3 cmH₂O). Inspiratory resistive loading for 2 min decreased potentiated Pdi_tw to 48.2 ± 2.8 cmH₂O (P < 0.01; Fig. 4). After 20 min of resting breathing, potentiated Pdi_tw was 44.7 ± 2.7 cmH₂O, which was less than the values at baseline (P < 0.01) and immediately after conclusion of 2 min of loading (P < 0.01; Fig. 4). After the 4-min period of loading, potentiated Pdi_tw was 43.4 ± 2.3 cmH₂O, which was less than that at baseline (P < 0.01). After 20 min of resting breathing, potentiated Pdi_tw decreased further to 41.3 ± 2.4 cmH₂O, which was less than the Pdi_tw values recorded at baseline (P < 0.01) and just before 4 min of loading (P < 0.05). At the conclusion of the task-failure protocol, potentiated Pdi_tw was 40.0 ± 2.2 cmH₂O, which was less than that recorded at baseline (P < 0.01). After 20 min of resting breathing, potentiated Pdi_tw decreased to 38.0 ± 2.1 cmH₂O, which was less than baseline (P < 0.01). Interestingly, the percent fall in the potentiated Pdi_tw was greater than that recorded for the nonpotentiated Pdi_tw (P < 0.001). After ~20 h, potentiated Pdi_tw increased to 48.2 ± 2.6 cmH₂O (n = 6), which was greater than the value recorded ~20 h earlier but was still less than baseline (P < 0.01; Fig. 4). As was the case for nonpotentiated Pdi_tw, the amplitudes of the right and left CDAPs of potentiated Pdi_tw were the same throughout this ~20-h period (Fig. 5), indicating a constant level of diaphragmatic recruitment.

CO₂ rebreathing. PET_CO2 values immediately before each CO₂ rebreathing challenge did not differ among the five experimental conditions (Table 1). The total time of rebreathing and the rate of PET_CO2 increase were similar at baseline, after the three periods of inspiratory resistive loading, and ~20 h later (Table 1).

View larger version (19K): [in this window] [in a new window]   Fig. 6.   Tidal volume (VT), breathing frequency (f), and minute ventilation (E) during CO2 rebreathing plotted against end-tidal PCO2 (PETCO2). Neither VT nor f was affected by preceding loading period sustained for 2 min, 4 min, and until task failure (n = 10). At 20 h after task-failure protocol, VT and f response to CO2 was the same as before loading (n = 6). E response to CO2 increased after 4th min of loading, and after ~20 h it was still greater than that recorded before loading. Values are means ± SE.

View larger version (21K): [in this window] [in a new window]   Fig. 7.   Inspiratory pressure changes for esophageal pressure (Pes), gastric pressure (Pga), and transdiaphragmatic pressure (Pdi) plotted as a function of PETCO2. Recordings were obtained at baseline, after inspiratory resistive loading for 2 min, 4 min, and until task failure (n = 10), and ~20 h later (n = 6). Pga contribution to tidal breathing was progressively decreased in relation to duration of preceding loading (P < 0.025), whereas Pes contribution to tidal breathing was not affected by preceding duration of resistive loading. Pdi tended to progressively decrease in proportion to duration of preceding loading, but this pattern did not reach statistical significance. These patterns were still present ~20 h after conclusion of loading to task failure. Values are group means ± SE.

View larger version (18K): [in this window] [in a new window]   Fig. 8.   Relationship between transpulmonary pressure (PL) and Pga before and after inspiratory resistive loading sustained for 2 min, 4 min, and until task failure at different levels of PETCO2. PL and Pga values (which represent start and end of inspiratory PL and Pga values, respectively) were obtained during CO2 rebreathing performed 20 min after each period of loading (n = 10) and at ~20 h after conclusion of loading to task failure (n = 6). Before fatigue there was a consistent increase in slope of PL-Pga relationship with increase in PETCO2, indicative of an increase in rib cage muscle contribution to tidal breathing. This pattern was more pronounced after 3 periods of loading starting at 51-55 Torr PETCO2. Before fatigue, at beginning of inspiration there was a progressive decrease in PL and increase in Pga with increase in PETCO2, which is consistent with expiratory muscle recruitment. This pattern was present and more pronounced after loading. A rest period of ~20 h after task failure was not sufficient to normalize these alterations in respiratory muscle recruitment. Values are means ± SE.

	DISCUSSION

Top Abstract Introduction Methods Results Discussion References

Inspiratory resistive loading sustained for periods of increasing duration decreased diaphragmatic contractility in proportion to the duration of loading, ever before task failure (Figs. 3 and 4). In addition, the pattern of respiratory muscle recruitment during an increase in ventilatory demand (CO₂ rebreathing) was altered as the duration of loading was increased (Figs. 7 and 8), and a rest period of ~20 h after task failure was not sufficient to normalize this alteration in respiratory muscle recruitment or the fatigue-induced changes in diaphragmatic contractility.

Sensations of air hunger and inspiratory effort. The sensation of inspiratory effort at the 4th min of the task-failure protocol was greater than at the same time in the preceding 4-min protocol (Fig. 2). Sense of effort during forceful muscle contractions is thought to be due to the magnitude of the central motor command to contracting muscles: the greater the command signal, the greater the neuronal corollary discharge being transmitted from the motor regions to sensory areas (i.e., subcortical postcentral areas and dorsal prefrontal cortex) (10). These corollary discharges are thought to be experienced as an increase in the sense of effort for a given load (13). We speculate that the loads sustained for 2 and 4 min had already initiated a contractile fatigue process in the diaphragm. Progressive diaphragmatic fatigue, in turn, required increased diaphragmatic recruitment (i.e., an increase in the magnitude of the central motor command signal) to maintain the target pressure during loading to task failure, i.e., sense of effort was greater at 4th min of the taskfailure protocol than at 4th min of 4-min protocol. Moreover, Ward et al. (32) reported that when the diaphragm becomes fatigued during resistive loaded breathing, the sensation of inspiratory effort increases in association with the increases in activity of the rib cage and sternomastoid muscles. Therefore, inspiratory muscles other than the diaphragm may have gradually increased their participation over the time of loaded breathing, accounting for the increasing inspiratory effort sensation in the present study.

Inspiratory resistive loading and Pdi_max. Pdi_max decreased after loading sustained for only 2 min and continued to decrease further after loading for 4 min and after loading to task failure (Fig. 3). At ~20 h of rest after the task-failure protocol, Pdi_max had not returned to the preloading value. The decrease in Pdi_max immediately after each of the three periods of loading could have resulted from central fatigue and/or contractile fatigue. Because the extent of voluntary activation of the diaphragm was not measured, we can only speculate on the likelihood of central fatigue. McKenzie et al. (21) measured the degree of voluntary diaphragmatic activation in healthy subjects during Pdi_max maneuvers recorded 10 s after each of three sets of 10 maximal expulsive efforts. Pdi_max decreased progressively, a maximal decrement of ~22% of the initial value, but the degree of diaphragmatic voluntary activation did not change over time. This suggests that contractile fatigue, rather than central fatigue, was probably responsible for the decrease in Pdi_max in our subjects.

Inspiratory resistive loading and Pdi_tw. A decrease in nonpotentiated Pdi_tw was observed after inspiratory resistive loading for 4 min and to task failure (Fig. 4). These results suggest that contractile fatigue is not an all-or-none phenomenon that occurs only when the respiratory muscles are loaded to task failure, as reported by Eastwood et al. (9). Our hypothesis is further supported by the decrease in potentiated Pdi_tw and Pdi_max after as little as 2 min of loading (Figs. 3 and 4). Also, the diaphragmatic contribution to tidal breathing in the presence of increased ventilatory demands (CO₂ rebreathing) tended to decrease after 2 min of loading (see below; Fig. 7). The early onset of contractile fatigue is further supported by the additional decreases in Pdi_max, potentiated Pdi_tw, and diaphragmatic contribution to tidal breathing during CO₂ rebreathing (see below) after 4 min of resistive loading.

Inspiratory resistive loading and respiratory muscle recruitment. To study the recruitment patterns of respiratory muscles in response to increased ventilatory demands, the subjects rebreathed CO₂. Juan et al. (12) showed that CO₂ per se depresses diaphragmatic contractility, but the exposure in their study was longer (10-15 min) than in the present study (3-4 min). Yan et al. (35) did not observe a decrease in the amplitude of Pdi_tw during CO₂ rebreathing, indicating that a brief period of hypercapnia does not impair diaphragmatic contractility. Therefore, the changes we observed in muscle recruitment are unlikely to be due to a direct depressant effect of CO₂ on the diaphragm per se but rather a consequence of the preceding periods of resistive loading. Despite what we interpret as progressively greater degrees of diaphragmatic fatigue (Figs. 3 and 4), our subjects showed no decrease in the ventilatory response to CO₂ (Fig. 6); this was achieved by even greater recruitment of rib cage and expiratory muscles. The alterations in the pattern of respiratory muscle recruitment had not reverted to normal after ~20 h of rest after task failure.