Stabilising function of the human diaphragm in response to involuntary augmented breaths induced with or without lower‐limb movements

New Findings What is the central question of this study? Is the stabilising function of the diaphragm altered differentially in response to involuntary augmented breaths induced with or without lower‐limb movements? What is the main finding and its importance? At equivalent levels of ventilation, the diaphragm generated higher passive pressure but moved significantly less during incremental cycle ergometry compared with progressive hypercapnia. Diaphragm excursion velocity and power output did not differ between the two tasks. These findings imply that the power output of the diaphragm during stabilising tasks involving the lower limbs may be preserved via coordinated changes in contractile shortening. Abstract Activity of key respiratory muscles, such as the diaphragm, must balance the demands of ventilation with the maintenance of stable posture. Our aim was to test whether the stabilising function of the diaphragm would be altered differentially in response to involuntary augmented breaths induced with or without lower‐limb movements. Ten healthy volunteers (age 21 (2) years; mean (SD)) performed progressive CO2‐rebreathe (5% CO2, 95% O2) followed 20 min later by incremental cycle exercise (15–30 W/min), both in a semi‐recumbent position. Ventilatory indices, intrathoracic pressures and ultrasonographic measures of diaphragm shortening were assessed before, during and after each task. From rest to iso‐time, inspiratory tidal volume and minute ventilation increased two‐ to threefold. At equivalent levels of tidal volume and minute ventilation, mean inspiratory transdiaphragmatic pressure (P¯di) was consistently higher during exercise compared with CO2‐rebreathe due to larger increases in gastric pressure and the passive component of P¯di (i.e., mechanical output due to static contractions), and yet diaphragm excursion was consistently lower. This lower excursion during exercise was accompanied by a reduction in excursion time with no difference in the active component of P¯di. Consequently, the rates of increase in excursion velocity (excursion/time) and power output (active P¯di × velocity) did not differ between the two tasks. In conclusion, the power output of the human diaphragm during dynamic lower‐limb exercise appears to be preserved via coordinated changes in contractile shortening. The findings may have significance in settings where the ventilatory and stabilising functions of the diaphragm must be balanced (e.g., rehabilitation).


INTRODUCTION
The diaphragm is the principal muscle of inspiration in humans (Grimby et al., 1976). During contraction, the diaphragm descends and flattens thereby expanding the lower ribcage and generating the negative pressure gradient that drives inspiratory flow. In addition to inspiration, the diaphragm muscle is involved in trunk stability during anticipatory postural adjustments (Hodges et al., 1997) and volitional upper-limb movements (Hodges & Gandevia, 2000a). Specifically, diaphragm descent increases intra-abdominal pressure, which, in turn, stabilises the spine independently from the muscle's respiratory activity (Hodges & Gandevia, 2000b). Trunk stability is important for maximizing force generation and minimizing joint loads in all dynamic activities, ranging from simple functional tasks to complex athletic movements (Hodges et al., 2004;Oddsson, 1988; see also Kibler et al., 2006). While previous research has provided valuable insight into the mechanisms for coordination of postural and respiratory functions of the diaphragm (Hodges et al., 1997;Hodges & Gandevia, 2000a, 2000b, the data were typically derived from repetitive arm movements (i.e., non-functional tasks). Whether the findings extend to dynamic lower-limb exercise is not yet clear.
The reference method for the assessment of diaphragm function is the measurement of transdiaphragmatic pressure (P di ), defined as the difference between pleural and abdominal pressure inferred from catheter-derived measures of oesophageal (P oe ) and gastric pressure (P ga ), respectively (Laveneziana et al., 2019). A major limitation, however, is that P di is merely an index of force output -it is a relatively poor index of contractile function (ATS/ERS, 2002;Laveneziana et al., 2019). The importance of this methodological limitation is emphasised by the finding that during dynamic lower-limb exercise (cycle ergometry) the diaphragm operates as a 'flow generator' (Aliverti et al., 1997); that is, the mechanical power of the diaphragm is mainly a function of shortening velocity rather than force/pressure. Imaging techniques, including fluoroscopy and magnetic resonance imaging (MRI), have been used to provide functional information regarding diaphragm shortening during static postural tasks (Kolar et al., 2010(Kolar et al., , 2012Vostatek et al., 2013). Unfortunately, fluoroscopy involves exposure to ionising radiation and MRI is influenced by motion artefacts. Thus, the applicability of these imaging techniques to dynamic exercise is limited.
In recent years, subcostal ultrasonography has emerged as a promising new technique for the quantification of diaphragm shortening. The technique can be used to quantify the mobility (excursion) of the crural diaphragm during tidal or forced inspiration (Laursen et al., 2021), and, if used in combination with established measures, could be used to provide estimates of diaphragm work (force [P di ] × distance [excursion]) and power output (P di × velocity [excursion/time]). Traditionally, ultrasonography has been used to evaluate diaphragm dysfunction and weaning outcome in critically ill patients (see Laursen et al., 2021). More recently, the technique has been used to better understand the associations between diaphragm mobility at rest and exertional dyspnoea, exercise limitation and health-related quality of life in patients with respiratory disease (Crimi

New Findings
• What is the central question of this study?
Is the stabilising function of the diaphragm altered differentially in response to involuntary augmented breaths induced with or without lower-limb movements?
• What is the main finding and its importance?
At equivalent levels of ventilation, the diaphragm generated higher passive pressure but moved significantly less during incremental cycle ergometry compared with progressive hypercapnia. Diaphragm excursion velocity and power output did not differ between the two tasks.
These findings imply that the power output of the diaphragm during stabilising tasks involving the lower limbs may be preserved via coordinated changes in contractile shortening. et al., 2018;Santana et al., 2019). Diaphragm ultrasonography can be performed repeatedly without adverse effects (see 'Technical considerations'), and, as such, has the potential to provide new insight into the stabilising function of the diaphragm during dynamic lower-limb exercise.
The aim of the present study was to test the hypothesis that the stabilising function of the diaphragm would be altered differentially in response to involuntary augmented breaths induced with or without lower-limb movements in healthy humans. To test this hypothesis, we used subcostal ultrasonography in combination with conventional measures (P di ) to quantify the dynamic contractile function of the human diaphragm in response to progressive hypercapnia versus incremental cycle ergometry.

Ethical approval
The study received ethical approval from the Brunel University

Participants
Ten healthy, recreationally active adults (five male and five female) with no history of smoking or cardiorespiratory disease were recruited.
Additional inclusion criteria were: age 18−30 years, body mass index 18.5−30.0 kg/m 2 and pulmonary function values above the lower limit of normal (see below). Participants abstained from strenuous exercise and alcohol for at least 24 h, caffeine for 12 h and food for 3 h before testing.

Experimental overview
Participants visited the laboratory on two occasions. The first visit was for screening, assessment of baseline characteristics and familiarisation with the experimental protocols (including instrumented CO 2 -rebreathe and exercise tests). The second visit was the experimental trial, which consisted of involuntary augmented breaths induced by progressive hypercapnia (CO 2 -rebreathe) and incremental cycle ergometry (exercise). The two tasks were performed in a pre-determined order (CO 2 -rebreathe, 20 min rest, exercise) to negate any potential influence of exercise-induced respiratory muscle work or fatigue on ventilatory responses to CO 2 (Mador & Tobin, 1992 (Graham et al., 2019;Wanger et al., 2005). Maximum inspiratory and expiratory pressure (PI max and PE max ) were assessed using an electro-manometer (MicroRPM, CareFusion) and recorded as the highest 1-s plateau pressure generated during maximal inspiratory and expiratory efforts initiated from residual volume (RV) and total lung capacity (TLC), respectively (Green et al., 2002). Values were expressed in absolute units and as a percentage of predicted (Evans & Whitelaw, 2009;Quanjer et al., 2012;Stocks & Quanjer, 1995).

Diaphragm thickness, thickening and excursion
Thickness, thickening and excursion of the diaphragm were assessed using ultrasonography (Vivid 7 Pro, GE Medical, Horten, Norway). Participants adopted a semi-recumbent position. Absolute thickness of the right costal hemidiaphragm was measured at FRC and TLC using a high-frequency, linear transducer (4.0−11.0 MHz, 10L, GE Medical) as per established guidelines (Laursen et al., 2021). Briefly, the transducer was positioned perpendicular to the right anterior or mid-axillary line at the 8−10th intercostal space and angled medially so that the hemidiaphragm was imaged as a three-layered structure.
The intercostal space with the least lung obscuration at TLC was selected. Ultrasound cine-loops were recorded in B (brightness)mode and analysed offline as still-images (EchoPac v6.1, GE Medical).
Thickness of the right hemidiaphragm was measured, in triplicate, medially between two adjacent ribs between the inner edges of the pleural and peritoneal membranes. Two participants exhibited lung obscuration at TLC; for these individuals, diaphragm thickness was measured as close to TLC as possible. The relative thickening fraction of the right costal hemidiaphragm was calculated as: (thickness at TLC -thickness at FRC)/thickness at FRC (Laursen et al., 2021). Maximum diaphragm excursion, defined as the maximum excursion of the right crural hemidiaphragm from FRC to TLC, was measured using a lowfrequency, curvilinear transducer (2.4−5.0 MHz, 3.5C, GE Medical) as per established guidelines (Laursen et al., 2021). The transducer was positioned subcostally on the right mid-clavicular line and angled cranially to image the right crural hemidiaphragm from the costophrenic angle to the inferior vena cava. Excursion was measured using angle-independent (anatomic) M (motion)-mode (additional details below).

2.5
Visit 2 2.5.1 CO 2 -rebreathe task The CO 2 -rebreathe protocol was similar to Read's hyperoxic, hypercapnic rebreathe method (Read, 1967 followed by the instruction to 'close your eyes, relax and breathe as needed' . When end-tidal partial pressure for CO 2 (P ETCO 2 ) reached 55 mmHg, the rebreathe circuit was opened to ambient air and the participant continued to breathe for 3 min (recovery). The laboratory was kept silent before, during and after the rebreathe task to prevent fluctuations in arousal level and the influence of participantexperimenter interaction (Homma & Masaoka, 2008;Spengler & Shea, 2001 (Wasserman et al., 2012, p.158). This duration was chosen on the basis that it was near the expected duration for CO 2 -rebreathe, yet still at the lower end of the range expected to elicit maximum physiological responses (Buchfuhrer et al., 1983). The exercise was terminated when pedal cadence dropped below 50 rpm for more than 5 s despite verbal encouragement. The participant continued to breathe on the mouthpiece for a further 3 min for the collection of recovery data.

Perceptual ratings
At the end of CO 2 -rebreathe and exercise, participants were asked to rate their dyspnoea intensity ('How strong/intense is your breathing discomfort?') using Borg's modified category ratio scale (CR-10; Borg, 1998). At the end of exercise, participants were also asked to rate their intensity of leg discomfort using the CR-10 scale. The scale was anchored at 0 ('no breathing [leg] discomfort') and 10 ('the most severe breathing [leg] discomfort ever experienced or imagined'). Finally, participants were asked to verbalise their main reasons for exercise cessation (breathing, legs, combination of both, other reasons).

Ventilatory and pulmonary gas-exchange indices
Ventilatory and pulmonary gas-exchange indices were assessed breath-by-breath using an online system that comprised a turbine flow meter and O 2 and CO 2 gas analysers (Oxycon Pro, Jaeger, Viasys Healthcare, Hoechberg, Germany). The flow meter and gas analysers were calibrated immediately before each task according to the manufacturer's recommendations. Raw signals from the online system were input into a data acquisition system (Micro1401-2 and Spike2, CED, Cambridge, UK) using an external device (DAQ-30A16, Eagle Technology, Cape Town, South Africa) for the offline determination of inspiratory minute ventilation (V I , BTPS), inspiratory tidal volume (V TI , BTPS), respiratory frequency (ƒ R ), inspiratory time (T I ), and inspiratory duty cycle (T I /T TOT ) (see below). Metabolic and gasexchange indices (V O 2 , P ETCO 2 , etc.) were taken directly from the online system.

Intrathoracic pressures
Oesophageal and gastric pressure (P oe and P ga ) were measured using two balloon-tipped catheters (CooperSurgical, Berlin, Germany).
Placement of the catheters was preceded by the administration of 2% lidocaine gel to the nasal mucosa. Each catheter was passed via the nares and swallowed. Once in the stomach, each catheter was connected to a low-range differential pressure transducer (DP45-3, Validyne Engineering, Northridge, CA, USA) that was calibrated across the physiological range using a digital manometer (C9553, JMW, Harlow, UK). The balloons were filled with 1 ml (P oe ) or 2 ml (P ga ) of air using a 10 ml glass syringe. The oesophageal catheter was withdrawn from the stomach until a negative deflection in P oe was observed upon inspiration. The catheter was withdrawn a further 10 cm so that the distal end was situated in the lower one-third of the oesophagus. A dynamic occlusion test was performed to confirm the oesophageal catheter was in the correct position (Baydur et al., 1982), after which, both catheters were taped in position at the nose. The volume of air in each catheter was checked at regular intervals to ensure no leaks had occurred. Analogue pressure signals were amplified (CD280, Validyne), digitised (Micro1401-2, CED), recorded online at 200 Hz (Spike2, CED), then down-sampled to correspond with the ventilatory signals (100 Hz). Instantaneous transdiaphragmatic pressure was calculated online (P di = P ga − P oe ).

Diaphragm shortening
Images of crural diaphragm shortening were obtained using a lowfrequency, convex-array ultrasound transducer (2.4−5.0 MHz, 3.5C, GE Medical) which facilitated a wide field of view with adequate framerate of capture (40−60 frames/s). The transducer was positioned subcostally on the right mid-clavicular line, as per established procedures (Laursen et al., 2021). Here, the right hemidiaphragm dome was imaged at the interface of the visceral pleura and air-filled lung. Beam penetration depth was adjusted so that the hyperechoic diaphragm at RV was always within the field of view (∼200−250 mm). The inferior vena cava was used as an anatomical landmark to ensure the region of interest and the insonation angle were kept constant. To optimise lateral resolution, one focal point was set at the diaphragm position at relaxation volume. The site of transducer placement was clearly marked with indelible ink to ensure replication during subsequent scans.
To minimise movement of the transducer, a step stool was used to rest the sonographer's arm and wrist on their knee during imaging. In addition, two ECG electrodes (3M, Bracknell, UK) were placed on the right and left mid-axillary lines so that the ultrasound system could monitor changes in ribcage volume via changes in distance between the two electrodes. The waveform signal of ribcage volume shown online during scanning and offline during analysis was used to distinguish between inspiration and expiration (see below).
To ensure consistency, the same experimenter (C.I.) performed all assessments.
Ultrasound cine-loops were recorded in B-mode, then analysed offline using angle-independent (anatomic) M-mode ultrasound (EchoPac v6.1, GE Medical). The dynamic cursor was orientated along the true axis of diaphragm movement (Orde et al., 2016). 2.5.7 Data processing and time-matching The start and finish of each breath were marked at points of zero flow. Anomalous breaths (e.g., swallows, coughs, sighs or breaths not crossing zero flow) were manually excluded. Inspired and expired volumes were derived via numerical integration of flow. Tidal pressures were expressed as mean inspiratory pressures (P di ,P ga andP oe ). For instance,P di was calculated as Σ n 0 (P di /n), where P di (0) was the absolute P di at the onset of inspiration and n was the number of data points during inspiration (Barnard & Levine, 1986). The active component ofP di (P di,a ) was calculated by subtracting the lowest pressure during any given respiratory cycle from the instantaneous pressure. The passive component (P di,p ) was calculated by subtractingP di,a fromP di . Accordingly,P di,a was the pressure required to accomplish inspiration (i.e., dynamic contractions), whereasP di,p was the pressure required to brace and stabilise the trunk (i.e., static contractions). The ratioP di /P oe (active pressures) was used to estimate the pressure contribution of ribcage muscles relative to that of the diaphragm during inspiration (Yan & Kayser, 1997). End-expiratory and end-inspiratory oesophageal pressure (EEP oe and EIP oe ) were used as indexes of end-expiratory and end-inspiratory lung volume (EELV and EILV), respectively.
To ensure the respiratory cycles within each ultrasound cine-loop were correctly matched with the breath-by-breath ventilatory and pressure data, particular attention was paid to the timing of ultrasound acquisition. Specifically, 15 s ultrasound cine-loops were recorded twice at rest, then every 30 s during hyperpnoea and recovery. All breaths recorded within a cine-loop were identified in the dataacquisition system and averaged over 15 s. Since all participants achieved a test duration of at least 1.75 min, data up to and including this time point (iso-time) were used for analysis. The final 15 s of hyperpnoea was also recorded, and defined as the peak response.

F I G U R E 1
Representative ultrasound images of the right hemidiaphragm during the final 15 s of CO 2 -rebreathe (P ETCO 2 , 55 mmHg; V I 32 l/min). B-mode image (top) was initially used to obtain the best diaphragmatic delineation, with the inferior vena cava (IVC) used as an anatomical landmark to ensure consistency in the positioning of the transducer. In anatomic M-mode (bottom), the diaphragm excursion (cm), excursion velocity (slope, cm/s) and excursion time (s) were measured for the estimation of diaphragm work and power output (see text for details). Vertical arrows indicate the beginning and end of a diaphragm contraction

Statistics
Statistical analysis was performed using dedicated software (SPSS Greenhouse-Geisser correction was applied when sphericity was violated. Effect sizes were reported as partial eta squared ( 2 p ), and interpreted as small ( 2 p = 0.01), medium ( 2 p = 0.06) or large ( 2 p = 0.14) (Lakens, 2013). The rate of change (i.e., slope) in diaphragm shortening (excursion and excursion velocity) versus ventilatory parameters (V TI and V TI /T I ) was determined for group mean data using linear regression. Slopes for the two tasks were compared using the calculations provided by Zar (2014). Descriptive data are reported as mean (SD). Statistical significance was set at P ≤ 0.05.

Participant characteristics
Participant characteristics are shown in Table 1. As expected, males were taller and heavier than females and exhibited larger chest dimensions. Static and dynamic lung volumes and maximum expiratory pressure were higher in males compared with females. However, all participants exhibited pulmonary function within normal limits.
Diaphragm thickness was similar for both sexes, whereas maximum diaphragm excursion was greater in males.

Ventilatory responses
Ventilatory and breathing pattern responses to CO 2 -rebreathe and exercise are shown in Figure 2 (see also Table 2

Respiratory pressure responses
Intrathoracic-pressure responses to CO 2 -rebreathe and exercise are shown in Figures 3-5 (see also Table 3). Despite similar ventilatory responses (see above),P di was consistently higher during submaximal exercise compared with CO 2 -rebreathe (∼44%; Figure 3a). These differences inP di were largely due to greater increases inP ga versus P oe (Figure 3b,c). As expected,P di at peak was higher for exercise due to the greater ventilatory demand. TheP di /P oe ratio was stable over time and similar for both tasks (Figure 3d), thereby indicating no differences in the inspiratory pressure contribution of ribcage muscles relative to that of the diaphragm. EEP oe did not differ from rest in either task (Figure 4), thereby indicating no changes in end-expiratory lung volume due to expiratory muscle recruitment or dynamic lung hyperinflation. In contrast, EIP oe (a surrogate for end-inspiratory lung volume) decreased at a faster rate and tended to be less negative (not significant) throughout exercise versus CO 2 -rebreathe ( Figure 4).    responses'), diaphragm work (P di,a × excursion) and diaphragm power (P di,a × excursion velocity) also did not differ between the two tasks F I G U R E 2 Ventilatory and breathing pattern responses to CO 2 -rebreathe and exercise. Data are means (SD) for 10 participants. See text for abbreviations

F I G U R E 3
Intrathoracic-pressure responses to CO 2 -rebreathe and exercise. Data are means (SD) for 10 participants. See text for abbreviations ( Figure 7). Due to the higher ventilatory demand at peak exercise (relative to CO 2 -rebreathe), the measures of excursion, velocity and power were also higher at peak (Figures 6 and 7) and throughout recovery. Positive linear relationships were noted for V TI versus diaphragm excursion and V TI /T I versus excursion velocity ( Figure 8); the mean rates of rise (slopes) did not differ between the two tasks, indicating that the blunting of diaphragm excursion during exercise was 'real' and not merely artefactual. Sex-disaggregated data are shown in Table 5. In general, the absolute values were larger in males but the relative changes were qualitatively similar for both sexes.

DISCUSSION
The aim of this study was to test whether the stabilising function of the human diaphragm would be altered differentially in response to involuntary augmented breaths induced with or without lowerlimb movements. We found that the diaphragm generated higher static (passive) pressure but moved significantly less (

Diaphragm excursion
It seems unlikely that the reduced excursion with lower-body exercise was attributable to a change in the diaphragm position at endexpiration because EEP oe (an index of EELV) was maintained close to resting levels throughout both tasks. This finding is consistent with Henke et al. (1988), who showed that active expiration was not elicited, and hence EELV was unchanged from rest, at similar levels of minute ventilation for cycle exercise and CO 2 -driven hyperpnoea when participants adopted a supine position. The most likely explanation for the lower excursion noted in the present study is a change in the diaphragm position at end-inspiration. Indeed, EIP oe (an index of EILV) tended to be less negative at equivalent time points throughout exercise compared with CO 2 -rebreathe. The reason for the apparent decrease in diaphragm excursion is not entirely clear.
One possibility is a change in the recruitment pattern of inspiratory muscles, although this seems unlikely because the relative contribution of ribcage muscles to inspiration (P di /P oe ) was similar for both tasks at comparable levels of tidal breathing. Another possibility is an increase in the movement of the ultrasound transducer or a change in its orientation during exercise, although, again, we think this is unlikely for several reasons. First, angle-independent (anatomic) M-Mode sonography was used to assess the excursion of the diaphragm.
With this approach, the cursor is positioned along the true axis of diaphragm excursion thereby leading to fewer orientation and translation errors (Orde et al., 2016). Second, the ultrasound transducer was anchored in relation to set anatomical landmarks to ensure the F I G U R E 5 Active and passive components of transdiaphragmatic pressure (P di,a andP di,p ) in response to CO 2 -rebreathe and exercise. Data are means (SD) for 10 participants. See text for abbreviations Data are means (SD) for five males (M) and five females (F).
transducer was in a consistent orientation during scanning. Finally, we showed that the mean rates of rise (slopes) for the relationship between V TI and diaphragm excursion (and V TI /T I vs. excursion velocity) were similar for both tasks. We are confident therefore that the reduction in diaphragm excursion with exercise was not merely a result of measurement error. Potential mechanisms for the observed reduction in diaphragm excursion with exercise are discussed below.

Respiratory-postural interactions
For equivalent levels of minute ventilation and tidal volume, the diaphragm generated higher force (P di ) yet moved significantly less (lower excursion) during exercise compared with CO 2 -rebreathe. By partitioning the active and passive components ofP di , it became evident that the higherP di during exercise was primarily due to an increase in the passive component (P di,p 2013) showed that diaphragm excursion was increased when the F I G U R E 8 Relationships between ventilatory and ultrasonographic measures of diaphragm shortening in response to CO 2 -rebreathe and exercise. Data are for 10 participants. Also shown are the linear regression equation (95% confidence interval for slope) and r 2 value for each task as well as the t-score (degrees of freedom) and P-value for the comparison of slopes. There were no differences in slopes between the two tasks lower limbs were loaded during isometric hip-flexion in healthy control subjects. Again, the apex and crural regions of the diaphragm were the predominant contributors to the resultant diaphragm excursion.
In all three studies, diaphragm excursion was assessed during a static postural task and compared with the response to tidal breathing at rest (Kolar et al., 2010(Kolar et al., , 2012Vostatek et al., 2013). In contrast, we evaluated diaphragm excursion during a task with high external validity (i.e., lower-limb cycle exercise) and compared the response to that induced by a task with low postural demand (i.e., be 'tonically' active for postural support, with phasic modulation for diaphragm contraction during inspiration (Hodges & Gandevia, 2000a, 2000b. Later, the same group expanded their findings by imposing a postural task (single-arm movement) during CO 2 -induced hyperpnoea (Hodges et al., 2001). When ventilatory demand increased, the 'tonic' activity (EMG) of the diaphragm and the phasic activity with arm movement was reduced or absent. The authors suggested that diaphragm postural inputs to the phrenic motoneurons may be 'gated' or 'occluded' when descending respiratory drive is increased. The findings imply that the coordination between postural and ventilatory demands is compromised when the chemical drive to breathe is increased. While Hodges et al. (2001) provide convincing evidence of the mechanism for coordination of postural and respiratory functions of the trunk muscles, it is not entirely clear whether their findings extend to lower-limb movements or, for that matter, any mode of dynamic exercise. Our study extends these previous findings by showing that the diaphragm contributes to trunk stability during dynamic lower-limb exercise through increased 'tonic' activity with restricted movement.

Maintenance of contractile function
To further characterise the stabilising function of the diaphragm, we calculated power output during inspiration as the product of the mean change in active P di (P di,a ), diaphragm excursion and 1/excursion time. Owing to concomitant reductions in inspiratory excursion and excursion time, excursion velocity (excursion/time) was similar for exercise and CO 2 -rebreathe. In combination with time-dependent increases inP di,a , the reductions in diaphragm excursion with exercise resulted in time-dependent increases in diaphragm work (P di,a × excursion) and power output that were similar for the two tasks.
Collectively, the results suggest that work and power of the diaphragm during stabilising tasks are maintained via coordinated changes in contractile shortening.
From rest to peak exercise, diaphragm power increased ∼12-fold due to a ∼4-fold increase in shortening velocity with only a ∼3-fold increase inP di,a . These magnitudes of increase are similar to those reported by others. For instance, Aliverti et al. (1997), using optoelectronic plethysmography (OEP) in combination with intrathoracic pressures, showed that diaphragm power increased up to 13-fold from quiet breathing to 70% of maximum work-rate during upright cycling. This increase was due to >6-fold increases in shortening velocity with a less than doubling of P di . Further, we observed a ∼5fold increase in diaphragm power during CO 2 -rebreathe that was due to a ∼3-fold increase in shortening velocity with a less than doubling of P di . This increase in power during CO 2 -rebreathe is similar to that reported by Romagnoli et al. (2004). Using OEP and intrathoracic pressures, those authors found a ∼5-fold increase in diaphragm power at a level of CO 2 -induced hypercapnia equivalent to that used in the present study (P ETCO 2 55 mmHg). While the authors did not quantify the absolute changes in diaphragm shortening velocity, they did note a decline in the pressure-to-velocity ratio throughout the task, albeit at much higher levels of hypercapnia (P ETCO 2 65−70 mmHg).
Thus, our data support the notion that the diaphragm may function primarily as a flow generator, rather than as a pressure generator, whereas the crural segment has a more important role in gastrooesophageal functions, such as retching and vomiting, and a significant role in providing a barrier during gastro-oesophageal reflux (Menezes & Herbella, 2017;Miller, 1990). Analysis of intra-breath profiles during progressive hypercapnia or hypoxia in dogs has shown that EMG activity of the crural segment precedes that of the costal segment during early inspiration and that the costal segment is fully shortened earlier during inspiration compared to the crural segment (Easton et al., 1993(Easton et al., , 1995. When comparing the mean shortening of both costal and crural segments during the course of an entire hypercapnic rebreathe trial, however, there are no significant differences in fibre shortening, shortening velocity or EMG activity between the two segments (Fitting et al., 1985). We are confident, therefore, that our ultrasonographic measures of crural diaphragm shortening provide a good representation of the contractile function of the entire muscle during involuntary augmented breathing.
Several steps were taken to minimise measurement error and bias. First, all participants were screened thoroughly to mitigate the influence of pathology or central obesity. Second, all participants completed a thorough familiarisation trial to accustom them to the rebreathe and exercise protocols. Third, great effort was made to control the external environment by keeping the laboratory silent before, during and after each task, thereby preventing fluctuations in arousal level and the influence of participant-experimenter interaction (Homma & Masaoka, 2008;Spengler & Shea, 2001). Fourth, established guidelines were followed in the acquisition of ultrasound data (Laursen et al., 2021). Fifth, all analyses were conducted by the same investigator and were performed consistently in line with clearly defined standards (Orde et al., 2016). Finally, the study used a repeated-measures design with a predetermined order of tasks (CO 2rebreathe followed by exercise) to avoid the potential influence of prior respiratory muscle work or fatigue on subsequent responses to hypercapnia (Mador & Tobin, 1992). Whilst it is conceivable that hypercapnia may alter the ventilatory response to subsequent ventilatory challenge (Griffin et al., 2012), it seems unlikely that such a response would be induced with short-term hypercapnia. Indeed, previous studies have shown that up to four repeated rebreathe tests over a 2 h period do not systematically alter the ventilatory response (Jensen et al., 2010;Sahn et al., 1977). A further consideration is that hypercapnic acidosis might elicit a reduction in diaphragm contractility, although, again, this seems unlikely on the basis that acute moderate hypercapnia has negligible effect on the P di response to phrenic nerve stimulation (Mador et al., 1997;Vianna et al., 1993). In short, it appears doubtful that our results can be attributed to interaction or carry-over effects.
Lastly, we did not consider biological sex to be relevant in the context of the present study and, therefore, did not design the study to detect sex-based differences. Nevertheless, we did ensure equal representation of male and female participants and considered the ultrasonographic data sufficiently novel to disaggregate by sex (see Table 5). In general, the absolute values for diaphragm excursion at rest were larger for males compared with females, but similar to values reported previously (Boussuges et al., 2020). Moreover, the relative changes within each task were qualitatively similar for males and females. Thus, our findings are likely to extend to both sexes.

Conclusion
We have shown that the contractile function of the human diaphragm during dynamic lower-limb exercise is relatively well preserved despite the possible restriction to inspiration that results from the increased requirement for postural stability. This supports the idea that the power output of the diaphragm during stabilising tasks involving the lower limbs is maintained via coordinated changes in contractile shortening. The findings further our understanding of the complex interplay between ventilatory and non-ventilatory functions of the human diaphragm and may have significance in settings where the ventilatory and stabilising functions of the diaphragm must be balanced (e.g., rehabilitation).

AUTHOR CONTRIBUTIONS
The experiments were performed at Brunel University London. Lee M.
Romer and Camilla R. Illidi conceived and designed the study. Lee M.
Romer and Camilla R. Illidi were involved in data collection and analysis.
Lee M. Romer and Camilla R. Illidi drafted the article and critically revised for important intellectual content. Both authors approved the final version of the manuscript and agree to be accountable for all aspects of the work in ensuring that questions relating to the accuracy or integrity of any part of the work are appropriately investigated and