Sex Differences in Durability Following Heavy Intensity Cycling Exercise in Trained Athletes.

E.P., M.C., S.P., C.G.B., and P.A. contributed to the conception and design of the work; E.P., P.S., E.S., L.B., H.K.W., M.C., C.F., A.M., S.C., B.M., W.P., and C.G.B. collected the data; E.P., P.S., E.S., P.S., M.C., C.F., A.M., S.C., B.M., W.P., and C.G.B. analyzed the data; E.P., M.C., S.P., C.G.B., and P.A. interpreted the data; E.P. drafted the article; E.P., E.S, P.S., M.C., S.P., C.G.B., and P.A. revised the article. All authors approved the final version of the article, agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. All persons designated as authors qualify for authorship, and all those who qualify for authorship are listed.

The authors have nothing to report.

Participants were fully informed of the aim and procedures of the study, including its risks and benefits, before signing the written informed consent form. All procedures conformed to the Declaration of Helsinki and were approved by the institutional ethical approval from the Northumbria University Health and Life Sciences Research Ethics Committee (submission reference: 2024‐6525‐6208). Using a small effect size ( η p 2 = 0.022 ) for a potential sex difference in the change in the power at ventilatory threshold (reliability statistic r  = 0.90) [ 30 ], with the parameters: α  = 0.05, 1 −  β  = 0.8, the a priori sample size calculation revealed that a minimum of 32 participants was required. Therefore, 32 healthy competitive cyclists and triathletes, including 16 females (mean ± SD age: 27 ± 4 years; stature: 164 ± 6 cm; mass: 56.8 ± 4.4 kg; V̇O 2peak : 51 ± 3 mL kg −1  min −1 ) and 16 males (age: 29 ± 6 years; stature: 181 ± 7 cm; mass: 75.9 ± 10.0 kg; V̇O 2peak : 58 ± 6 mL kg −1  min −1 ) were enrolled in the study after providing written informed consent. All participants were well‐trained (Performance Level 3 or above) [ 31 , 32 ], competed in cycling or triathlon races (at least one race in the last 3 months) and were familiar with ≥ 2 h exercise bouts. To ensure that the cardiovascular fitness of the two groups was similar, participants had to have a minimum V̇O 2peak (determined in the initial visit, see below) of 48 mL kg −1  min −1 for females and 55 mL kg −1  min −1 for males. These values correspond to equitable performance levels in sports science research [ 31 , 32 ]. Two additional participants were recruited but did not complete the experimental visit because they did not meet the V̇O 2peak inclusion criteria. Moreover, in order to be included in the study, all participants had to be aged between 18 and 40 years old, had to be free from cardiovascular, respiratory, or neurological diseases or illness as well as musculoskeletal injuries that could affect cycling performance in the last 6 months. Training characteristics (training volume and frequency) were collected through the initial screening questionnaire and are reported in Table  1 . Participant demographics and baseline incremental test results in males and females. Values are presented as mean ± SD. Note: Bold values indicate statistically significant differences between males and females ( p < 0.05). Abbreviations: Bla peak , peak blood lactate; GET, gas exchange threshold; HR peak , peak heart rate; P max , maximal power; RCP, respiratory compensation point; RER peak , peak respiratory exchange ratio; V̇CO 2peak , peak carbo dioxide production; V̇E peak , peak ventilation; V̇O 2peak , peak oxygen uptake. This was a multi‐site study, where participants were recruited and tested at Northumbria University ( n  = 11), Newcastle University ( n  = 12), and at the University of Pavia ( n  = 9) using the same methods and protocols. For female participants, both naturally menstruating females and hormonal contraceptive users were included. More specifically, naturally menstruating participants reporting a regular menstrual cycle duration (> 21 and < 35 days), and not using hormonal contraceptives for at least 6 months, had no menstrual cycle‐related irregularities (e.g., amenorrhea), or conditions known to affect the hypothalamic–pituitary–ovarian axis (e.g., polycystic ovarian syndrome, endometriosis, and pregnancy) [ 33 ] were included. Moreover, females who were taking a combined monophasic oral contraceptive pill (OCP) (containing a dose of 30 μg of EE) for at least 6 months were included, as well as females taking any other kind of continuous hormonal contraceptives (implants, intrauterine devices/coil, etc.). In our female cohort, n  = 11 were naturally menstruating females, n  = 1 was taking a combined monophasic pill, n  = 2 were using a Mirena coil, n  = 1 reported to have a Nexplanon contraceptive implant (progesterone only), and n  = 1 had a vaginal ring. Naturally menstruating females were tested during the early follicular phase of the menstrual cycle (both low estrogen and low progesterone levels, as from the onset of bleeding until day 5) whereas the oral contraceptive and the vaginal ring users were tested during the pill withdrawal phase and the removal period, respectively. During these time points, the hormonal environments are expected to be similar between these groups [ 34 ]. The other females reporting to have the Mirena coil or the Nexplanon contraceptive implant, which are progesterone only, were not tested during a specific phase as their hormonal environment remained constant during usage. Participants visited the laboratory on two occasions, separated by at least 72 h of rest (mean: 85 ± 7 h), and at the same time of day. The first visit involved a familiarization with neuromuscular measures, followed by a cycling incremental exercise test to the limit of tolerance. In the second visit, participants performed a fatiguing cycling exercise, involving 90 min of cycling in the heavy intensity domain at 110% of GET [ 35 ] followed by the same incremental test performed in visit 1. Before and immediately (commencing within 60 s) after the fatiguing exercise, maximal voluntary isometric contraction force (MVIC), voluntary activation (VA), and potentiated twitches (high‐frequency 100 Hz doublet, Db 100 ; low‐frequency 10 Hz doublet, Db 10 ; single stimulation, Q tw·pot ) of the quadriceps were assessed. During both incremental exercise tests, pulmonary gas exchange, heart rate (HR), and rating of perceived exertion (RPE) were recorded, and blood lactate (BLa) samples were collected. In addition to these measurements, during the 90 min of cycling in the heavy intensity domain, near‐infrared spectroscopy (NIRS) and electromyography were also recorded. All participants were asked to maintain their habitual diet throughout the study, recorded their diet 24 h prior to the first lab visit and replicated the same diet prior to the second visit, and were instructed to eat a high carbohydrate meal before attending each visit. Participants were also asked to arrive at the lab ≥ 2 h post prandial and in a hydrated state. Moreover, they were recommended to refrain from alcohol and vigorous exercise < 48 h prior to visits, consumption of nutritional supplements/ergogenic aids (e.g., sodium bicarbonate, nitrates) < 24 h prior to visits and consumption of caffeine < 6 h prior to each visit. This visit began with participants being familiarized with the neuromuscular stimulation techniques. After the determination of the femoral nerve stimulation intensity, participants performed a series of warm up contractions increasing from 50% perceived effort to 90%. Hereafter, a neuromuscular assessment was performed (see full details below), before participants moved to the cycle ergometer (Lode Excalibur, Lode B.V., Groningen Country, NL) to perform a modified step‐ramp protocol [ 35 ], from that developed by Iannetta et al. [ 36 ]. Resting measurements of pulmonary gas exchange and HR were recorded for 3 min, and a blood lactate sample was taken. Then, the step‐ramp protocol began with 2 min of cycling at 20 W, followed by 6 min of cycling at 1.5× body mass (males) or 1.3× body mass (females). Thereafter, participants completed 4 min of cycling at 20 W before commencing the ramp test, which increased the work rate by 20 W· min −1 for females, or 25 W· min −1 for males. The ramp test continued until task failure, defined as the point at which cadence decreased by 10 rpm from the participant's self‐selected cadence (75–95 rpm) for more than 5 s despite strong verbal encouragement. This visit began with a neuromuscular assessment evaluation (see below). Then, participants moved to the cycle ergometer where recordings of resting pulmonary gas exchange, near‐infrared spectroscopy, and electromyography were taken and a resting blood lactate sample was collected. Participants began with a warm‐up consisting of a 5‐min stage at 75% of the power output associated with GET, after which the power output instantly increased to the target power output for the 90 min of heavy‐intensity exercise at 110% GET. Participants were instructed to maintain their preferred cadence (between a range of 80 and 90 rpm) and they were given the choice of listening to music during the 90‐min trial and having a visible clock showing the time remaining. Participants were also allowed to drink water ad libitum but not consume any food or calorie‐containing drinks during the test. Immediately (commencing within 60 s) after the completion of the fatiguing task, participants repeated the neuromuscular assessments and then performed the same step‐ramp protocol as the first visit. Following the step‐ramp protocol, participants laid supine, whilst blood flow was occluded for 5 min to allow a physiological calibration of NIRS signals (see below). During the fatiguing task, heart rate, near‐infrared spectroscopy, and electromyography were recorded continuously, whereas breath‐by‐breath gas exchange measures were taken for the first 15 min, and then for 5 min every 15 min thereafter. The metabolic cart was recalibrated between measurements to ensure accurate and reliable gas exchange measurements. Ratings of perceived exertion (6–20 scale) [ 37 ] and blood lactate samples were collected every 15 min. The full protocol is displayed in Figure  1 . Graphical representation of the experimental procedure performed in visit 2. The y axis represents the exercise intensity during each stage of the protocol whereas the x axis shows the timeline of the protocol. The chair figure represents a neuromuscular assessment (consisting of MVICs and potentiated twitches at rest). The colored background highlights the cycling stages, whereas the arrows indicate time points for physiological and perceptual assessment (breath by breath gas exchanges, blood lactate, heart rate, electromyography, and rate of perceived exertion). Adapted from Colosio et al. [ 38 ]. During all visits, pulmonary gas exchange and ventilation were measured breath‐by‐breath using an online system (Vyntus CPX, Jaeger, CareFusion, Germany), with minute ventilation (V̇E), oxygen consumption (V̇O 2 ), carbon dioxide production (V̇CO 2 ) and respiratory exchange ratio (RER) being quantified. Before each test, the analyzers for oxygen and carbon dioxide were calibrated using ambient air and a gas of known concentration (15% O 2 and 4.97% CO 2 ). Ventilatory volumes were calibrated using a digital turbine transducer at high (2 L s −1 ) and low (0.2 L s −1 ) flow rates. Following warm up contractions, baseline neuromuscular assessments consisted of five MVICs separated by 60 s. During the final three MVICs, a high‐frequency doublet electrical stimulation [100 Hz, 10‐ms interstimulus interval (sDb 100 )] was superimposed to the contraction at peak force. Subsequently, once the muscle was relaxed, a series of potentiated high (100 Hz) and low‐frequency (10 Hz, 100‐ms interstimulus interval, Db 10 ) doublet stimulations, as well as a single twitch (Q tw·pot ), were delivered. This assessment protocol was used to quantify MVIC force, voluntary activation (VA), and contractile function through Db 100 , Db 10:100 and Q tw.pot [ 38 ]. At the end of the fatiguing task, in the interest of time the neuromuscular assessment was repeated but with three MVICs instead of five, as re‐potentiation of twitches was not required [ 39 ]. Electrical stimuli (200 μs duration) were delivered to the femoral nerve via 32 mm‐diameter surface electrodes (CF3200; Nidd Valley Medical, North Yorkshire, UK) using a constant‐current stimulator (DS7AH, Digitimer, Welwyn Garden City, Hertfordshire, UK). The cathode was placed high in the femoral triangle over the nerve, and the anode was positioned midway between the greater trochanter and the iliac crest. The cathode was repositioned, if necessary, to the point that elicits the largest quadriceps twitch amplitude (Q tw ). Stimulations began at 20 mA and increased by 20 mA until a plateau in Q tw occurred. This value was then increased by 30% to ensure supramaximal stimulation was delivered during the protocol. Mean stimulus intensity was not different between males (309 ± 108) and females (260 ± 152 mA, p  = 0.360). For the neuromuscular assessments, participants were seated on a custom‐built chair, with force (N) measured using a calibrated load cell (MuscleLab force sensor 300, Ergotest technology, Norway or SML load cell, Interface, Scottsdale, AZ). The load cell was attached to the participant's dominant leg, 2 cm superior to the ankle malleoli, using a non‐compliant cuff. The load cell height was adjusted to ensure a direct line with the applied force for each participant. Participants sat upright with knee and hip angles kept at 90° flexion. The chair was placed ~1 m adjacent to the cycle ergometer to facilitate movement and save time for the post assessment. Force was sampled continuously (1000 Hz), and acquired for offline analysis (Spike 2, Cambridge Electronic Design, Cambridge, UK or LabChart 8, ADInstruments, Bella Vista, Australia). EMG signals were recorded continuously at 2000 Hz throughout both visits using wireless sensors (10 mm inter‐electrode distance; Trigno Avanti, Delsys, MA, USA or Pico EMG, Cometa, Bareggio, Italy). Sensors were placed over the participant's dominant leg on Rectus Femoris (RF), Vastus Lateralis (VL), and Vastus Medialis (VM) consistent with SENIAM guidelines [ 40 ]. Prior to placement, the skin‐electrode contact area was shaved, abraded, and cleaned using a 70% IPA alcohol wipe (FastAid, Robinson Healthcare, Workshop, UK). The raw EMG signals were amplified (gain ×150) and, using a custom script (MATLAB 2024b, MathsWorks Inc., Natick, MA, US), filtered using a 20–450 Hz 4th order Butterworth bandpass filter. Signals were amplified: gain ×150 for EMG (Delsys Trigno EMG systems, Boston, MA, USA or Pico EMG, Cometa, Bareggio, Italy) and ×300 for force (CED 1902; Cambridge Electronic Design, Cambridge, UK or PowerLab 8/35, ADIntruments), bandpass filtered (EMG only: 20–450 Hz), digitized (EMG: 2 kHz; force: 5 kHz; CED 1401, Cambridge Electronic Design or PowerLab 8/35, ADInstruments or Load Cell Adapter, Delsys, Natick, MA), and analyzed offline (force: Spike2 v8, Cambridge Electronic Design, EMG: R2024; Mathworks Inc., Natick, MA). A wireless, portable, and continuous‐wave spatially resolved NIRS light photometer (Portamon, Artinis Medical Systems) was used to evaluate relative changes in oxy—(O 2 HbMb) and deoxy‐ (HHbMb) hemoglobin and myoglobin, as well as tissue saturation index (TSI: [O 2 Hb Mb]/total [Hb + Mb] × 100), sampled at 10 Hz. The probe was placed on the lower third of vastus lateralis muscle (~10 cm above the knee joint). The probe was held in place by an elasticised, tensor bandage and covered by an opaque, dark material to avoid motion and ambient light influences [ 41 ]. Before the start of both the incremental test and the fatiguing task, baseline TSI, O 2 HbMb and HHbMb were recorded over 3 min of rest. At the end of each visit, a prolonged ischemia (5 min) from femoral artery occlusion by cuff inflation with pressure above 120% of limb occlusion pressure, which was not different between sexes ( p  = 0.271), was performed by using a 13 × 85 cm rapid‐inflation pressure cuff (Delfi Medical Innovations Inc., Vancouver BC, Canada or E20, or Hokanson, Place Bellevue, WA, USA) placed proximally on the same thigh and attached to a cuff‐inflator. This procedure was used to perform the physiological calibration of the NIRS signal, and the relative changes in VL O 2 HbMb and HHbMb during the fatiguing task were expressed as percentage change from the minimum (0%) and the maximum (100%) value recorded during the physiological calibration. Blood lactate was sampled via capillary puncture technique with a 10 μL sample taken from the earlobe of each participant. Samples were immediately analyzed for the concentration of lactate (mmol L −1 , Biosen, EKF Diagnostics) and used to assess changes during the fatiguing task. All MVICs were recorded, and the single contraction with the greatest peak force (250‐ms window before the superimposed electrical stimulation) [ 42 ] used for further analyses. To quantify impairments to the central nervous system drive, knee extensors voluntary activation (VA) was calculated using the interpolated twitch technique and calculated from the equation [ 43 ]: VA (%) = (1 − [sDb 100 /Db 100 ] × 100), where sDb 100 is the amplitude of the high‐frequency doublet superimposed twitch force measured during MVIC, and Db 100 is the amplitude of the high frequency doublet twitch force assessed 2 s post‐MVC on relaxed muscle. Changes in skeletal muscle contractile properties were assessed by variations in the amplitudes of Q tw·pot , Db 100 , and Db 10:100 [ 44 ]. Peak‐to‐peak EMG amplitude elicited by single femoral nerve electrical stimulation was measured to assess sarcolemma excitability (M max ). The filtered EMG data were then root‐mean‐squared across a 200 ms moving window (EMG RMS ). The maximum EMG RMS elicited during the MVIC was expressed as a percentage of M max . For measurements of EMG during the incremental and fatiguing cycling tasks, the peak EMG RMS during each EMG burst (i.e., each VL, VM, and RF contraction per pedal rotation) was automatically identified and expressed as a percentage of both the pre‐exercise maximum EMG RMS as well as the pre‐exercise M max using a custom‐built script (R2024; Mathworks Inc., Natick, MA). Cycling EMG RMS was averaged at the same time points as gas exchange measurements and NIRS data. The rate of V̇O 2 , V̇CO 2 , and V̇E was recorded during both visits and was exported in 5 s intervals. Gas exchange threshold (GET) and respiratory compensation point (RCP) were visually, individually, and independently determined by two expert investigators after they agreed on the value set, using multiple gas exchange and ventilatory equivalent criteria through consideration of V̇O 2 , V̇CO 2 , end‐tidal partial pressures for CO 2 and O 2 , and V̇E/V̇CO 2 and V̇E/O 2 [ 45 , 46 ]. The power outputs associated with the GET and RCP were determined according to the methods outlined by Brownstein et al. [ 35 ], by linearly interpolating the V̇O 2 versus power output relationship after the V̇O 2 was left‐shifted by a time interval corresponding to the mean response time (MRT). To avoid the potential influence of the cycling trial in the heavy intensity domain on the V̇O 2 kinetics in the post ramp test, the correction of GET and RCP for MRT was performed in both PRE and POST ramp tests. Using the MRT from the moderate intensity step, the power output associated with GET was left‐shifted to account for the time delay between changes in muscle metabolism being reflected in pulmonary gas exchange [ 36 ]. V̇O 2peak was defined as the highest 30 s rolling average of V̇O 2 , and maximal power output (P max ) was considered as the last power output that each participant was able to complete during the ramp test before task failure. During the 90 min of fatiguing cycling, the variables V̇O 2 , V̇CO 2 , V̇E, RER as well as O 2 HbMb and HHbMb were averaged from 4 to 6 min after exercise commencement (in order to avoid phase I and II onset kinetics) and during the last 2 min of each 5 min time intervals (13–15 min, 28–30 min, 43–45 min, 58–60 min, 73–75 min, 88–90 min). Relative whole‐body energy expenditure (EE/kg) expressed in J s kg −1 was calculated according to the equation ( x ) = [(281.67 × V̇O 2 ) (L min −1 ) + (80.65 × V̇CO 2 ) (L min −1 )/kg] [ 47 ]. Gross efficiency (GE) (%) was calculated as Work rate (W)/Energy Expenditure (J s) × 100. Results are presented as means ± SD within the text and figures. Statistical significance was set at an α level of 0.05. Normality of the data was assessed by the Shapiro Wilk test, with no data requiring transformation. Assumptions of sphericity were explored and controlled for all variables with the Greenhouse–Geisser adjustment, where necessary. Sex comparisons for P max , V̇O 2peak , V̇CO 2peak , V̇E peak , HR peak , Bla peak , PO at GET and at RCP, GET and RCP (%V̇O 2peak ) and GET and RCP (%HR max ) were assessed with independent samples t ‐tests. Exercise‐induced changes in MVIC, VA, Db 100 , Db 10:100 , Q tw·pot , PO at GET, PO at RCP, P max , V̇O 2peak , BLa, HR, V̇O 2 , V̇CO 2 , V̇E, RER, EE, GE, EMG RMS , TSI%, O 2 HbMb, HHbMb and RPE were assessed with a two‐way repeated measures ANOVA. Each ANOVAs involved sex and time as the independent variables. Significant sex and time main effects or sex × time interaction effects were further explored using Bonferroni‐corrected pairwise comparisons. Partial eta squared ( η p 2 ) was calculated to estimate effect sizes, with values representing small ( η p 2 < 0.06 ), medium ( η p 2 ≥ 0.06 , < 0.14) and large ( η p 2 ≥ 0.14 ) effects. The statistical software package Prism 10 (GraphPad, Software Inc., San Diego, CA) was used to analyze data.

The variables recorded during the incremental ramp test performed in visit 1 are displayed in Table  1 . As expected, males showed greater values forV̇O 2peak (both absolute and relative to body mass), P max , V̇CO 2peak , V̇E peak , power outputs associated at GET and RCP (all p < 0.001). HR peak , Bla peak , V̇O 2 (%peak) and HR (%peak) associated at GET and RCP were not different between males and females ( p ≥ 0.053). MRT values for the ramp PRE were 12 ± 8 W and 9 ± 6 W for males and females respectively ( p  = 0.257), whereas for ramp POST were 8 ± 8 W and 7 ± 5 W for males and females respectively ( p  = 0.656). The average performance level [ 31 , 32 ] was similar between males and females (3 ± 1 vs. 3 ± 1, p  = 0.265). As expected, before HYV, males showed greater MVIC (630 ± 148 vs. 373 ± 75 N, p < 0.001), resting twitch responses for Db 100 (248 ± 81 vs. 170 ± 51 N, p  = 0.005), Db 10 (247 ± 76 vs. 164 ± 44 N, p  = 0.001), Q tw·pot (168 ± 58 vs. 108 ± 29 N, p  = 0.001) and M max (10.59 ± 5.81 vs. 6.94 ± 4.54 mV, p  = 0.042) than females. VA was not different between sexes (94 ± 4 vs. 91% ± 5%, p  = 0.136). Males performed HVY at a greater absolute intensity than females (247 ± 48 vs. 149 ± 23 W, p < 0.001) but at the same relative intensity, which corresponded to 110% of power output associated at GET. The power outputs for males and females were at 21% ± 7% and 17% ± 6% of the difference between GET and RCP, respectively ( p  = 0.07). Indeed, at the beginning of the fatiguing cycling trial (4–6 min) the V̇O 2 (%V̇O 2GET ) at the target intensity was not different between males (112.7% ± 7.4% V̇O 2GET ) and females (112.7% ± 10.3% V̇O 2GET , p  = 0.982). Of the whole sample of 32 participants, three male participants reached task failure at 60, 78, and 86 min of exercise and one female participant terminated the trial at 60 min of cycling. However, despite achieving task failure, they all completed the ramp test POST. Changes in incremental test results after HVY are displayed in Figure  2 . A time effect was found for P max (−10% ± 8%; F 1,30 = 51.83, p < 0.001, η p 2 = 0.63 ; Figure  2E ), for V̇O 2peak (−7% ± 7%; F 1,59 = 27.63, p < 0.001, η p 2 = 0.32 ; Figure  2F ), V̇CO 2peak (−15% ± 12%; F 1,59 = 50.84, p < 0.001, η p 2 = 0.46 ), BLa peak (−45% ± 20%; F 1,59 = 161.0, p < 0.001, η p 2 = 0.73 ), and RER peak (−9% ± 8%; F 1,59 = 39.8, p < 0.001, η p 2 = 0.40 ), with no effects of sex, or time × sex interactions ( p ≥ 0.072). A time × sex interaction ( F 1,59 = 4.04, p  = 0.049, η p 2 = 0.06 ) was found for V̇E peak , which decreased more in males compared to females (−18% ± 16% vs. −9% ± 10%, p  = 0.049). Incremental ramp test results after the 90 min of cycling in the heavy intensity domain. (A) Power output at gas exchange threshold (GET); (B) power output at respiratory compensation point (RCP); (C) Oxygen uptake (V̇O 2 ) at GET; (D) Oxygen uptake (V̇O 2 ) at RCP; (E) maximal power output (PO max ); (F) peak oxygen uptake (V̇O 2peak ). Male data are presented in blue and female data in red. Dots indicate individual participants and lines indicate the group mean ± standard deviation. A time × sex interaction was found for both PO at GET ( F 1,30 = 6.93, p  = 0.013, η p 2 = 0.19 ; Figure  2A ) and PO at RCP ( F 1,30 = 4.19, p  = 0.049, η p 2 = 0.12 ; Figure  2B ). Indeed, PO at GET decreased only in males (−16% ± 15%, p  = 0.001) but not in females (−2% ± 13%, p  = 0.460). Moreover, the decrease in PO at RCP was greater for males compared to females (−13% ± 10% vs. −6% ± 9%, p  = 0.049). A time effect was detected for V̇O 2 at GET ( F 1,59 = 5.82, p  = 0.019, η p 2 = 0.09 ; Figure  2C ), with no sex effect or time × sex interaction ( p  = 0.096). No effects of time, sex, and time × sex interaction were found for V̇O 2 at RCP ( p ≥ 0.169; Figure  2D ). Time effects and main effects of sex are displayed in Figure  2 . During HVY, a time effect was found for RPE ( F 5143 = 51.06, p < 0.001, η p 2 = 0.64 ), V̇E (%peak, F 6176 = 9.93, p < 0.001, η p 2 = 0.25 ), and fat oxidation ( F 6177 = 11.95, p < 0.001, η p 2 = 0.29 ) which all increased over time, as well as GE ( F 6175 = 6.86, p < 0.001, η p 2 = 0.19 ), which decreased over time; all without main effects of sex or time × sex interactions ( p ≥ 0.741). A time × sex interaction ( F 6168 = 3.12, p < 0.006, η p 2 = 0.10 ) was detected for HR, demonstrating greater increases in males at 60 ( p  = 0.042) and 90 min ( p  = 0.032; Figure  3A ). Similarly, RER demonstrated a time × sex interaction ( F 6175 = 3.84, p  = 0.001, η p 2 = 0.12 ), with greater decreases in females at 60 ( p  = 0.015) and 90 min ( p  = 0.016; Figure  3B ). A sex effect was found for BLa ( F 1,29 = 5.24, p  = 0.030, η p 2 = 0.15 ) with greater blood lactate concentrations in males than females at 30 ( p  = 0.039), 75 ( p  = 0.022) and 90 min of exercise ( p  = 0.018; Figure  3C ). A time × sex interaction ( F 6175 = 2.99, p  = 0.008, η p 2 = 0.09 ) was found for V̇O 2 (%peak PRE ), which increased more in males compared to females from 15 min of exercise onwards ( p ≤ 0.021; Figure  3D ). A time × sex interaction ( F 6175 = 3.14, p  = 0.006, η p 2 = 0.10 ) was detected for relative EE, showing greater increases in males than females ( p ≤ 0.001; Figure  3E ). Whereas a time × sex interactions ( F 6175 = 3.44, p  = 0.003, η p 2 = 0.11 ) was found for carbohydrate oxidation, which decreased in females from 45 min of exercise ( p ≤ 0.008) but not in males ( p ≥ 0.874). Time effects and main effects of sex are displayed in Figure  3 . Physiological responses during the 90 min of cycling in the heavy intensity domain. Changes in (A) heart rate (HR); (B) respiratory exchange ratio (RER); (C) blood lactate (Bla); (D) Oxygen uptake as %peak (V̇O 2 ); (E) energy expenditure (EE); (F) accumulated work as kJ·kg −1 during the 90‐min constant cycling in the heavy intensity domain. Dots indicate group mean and lines indicate the standard deviation. * Time effect; # time × sex interaction; $ sex effect; ϯ different between males and females. The changes in neuromuscular function are displayed in Figure  4 . A time effect was found for MVIC (−16% ± 12%; F 1,30 = 62.57, p < 0.001, η p 2 = 0.67 ; Figure  4A ), Db 100 (−15% ±13%; F 1,58 = 40.51, p < 0.001, η p 2 = 0.41 ), Db 10:100 (−22% ± 11%; F 1,58 = 135.9, p < 0.001, η p 2 = 0.71 ), Q tw·pot (−27% ± 13%; F 1,59 = 132.80, p < 0.001, η p 2 = 0.69 ), and VA (−6% ± 6%; F 1,59 = 24.11, p < 0.001, η p 2 = 0.29 ; Figure  4B ), without time × sex interaction effects ( p ≥ 0.096). Although M max did not show a main effect of time ( p  = 0.320), there was a sex effect ( F 1,56 = 11.99, p  = 0.001, η p 2 = 0.18 ) and a time × sex interaction ( F 1,56 = 11.99, p  = 0.001, η p 2 = 0.18 ), increasing in females (+19% ±35%, p  = 0.003) and not in males (−12% ±21%, p  = 0.082). Changes in neuromuscular function. (A) Changes in maximal voluntary isometric contraction (MVIC); (B) changes in potentiated twitches (Db 100 , Db 10:100 , Q tw·pot ) and voluntary activation (VA) by sex from PRE to POST. Dots indicate individual participants and bars indicate the group mean ± SD. For all parameters, the PRE‐POST decrease was statistically significant. The EMG RMS /MVC during HVY did not change in the VL over time during the prolonged cycling task (time effects p  = 0.560, initial values: 38.6% ± 18.6% MVC). Similar findings were reported for EMG RMS /M max for VL (2.4% ± 1.4% M max , time effect p ≥ 0.330). A time × sex interaction ( F 6147 = 11.91, p < 0.001, η p 2 = 0.33 ) was found for O 2 HbMb (%ischemia) which increased in females but not in males from 45 min of exercise (Figure  5A ). A time × sex interaction ( F 6147 = 3.30, p  = 0.005, η p 2 = 0.12 ) was found for HHbMb (%ischemia) which increased in males but not in females from 45 min of exercise (Figure  5B ). A time × sex interaction ( F 6139 = 7.04, p < 0.001, η p 2 = 0.23 ) was found for TSI (%) which decreased more in males than females from 15 min of exercise (Figure  5C ). Only a time effect ( F 6129 = 17.62, p < 0.001, η p 2 = 0.45 ) was found THb+Mb, which increased over time without sex or time × sex interaction effects ( p ≥ 0.079; Figure  5D ). Time effects and main effects of sex are displayed in Figure  5 . Indices of muscle oxygenation throughout the 90 min of cycling in the heavy intensity domain. (A) Oxyhaemoglobin and myoglobin (O 2 HbMb); (B) deoxyhaemoglobin and myoglobin (HHbMb); (C) tissue saturation index (TSI); (D) total hemoglobin and myoglobin (THbMb). Dots indicate group mean and lines indicate the standard deviation. * time effect; # time × sex interaction; $ sex effect; ϯ different between males and females.

The present study aimed to investigate durability and its physiological underpinnings in males and females following 90 min of constant‐load cycling in the heavy intensity domain. Females exhibited lesser reductions in the power outputs associated with submaximal thresholds, whereas data recorded at peak exercise decreased similarly in both sexes. Surprisingly, no sex differences were observed in the changes in neuromuscular function, which was impaired similarly in males and females at the end of the fatiguing task. During the 90 min of cycling in the heavy intensity domain, heart rate, relative V̇O 2 (%peak), and energy expenditure increased more in males than females, suggesting the development of a greater V̇O 2 drift and loss of efficiency as exercise proceeded. This was supported by the greater rise in HHbMb, suggesting elevated O 2 extraction in males compared to females. Sex differences were also observed in substrate oxidation, with females demonstrating a lesser reliance on carbohydrates in the latter stages of exercise. These physiological sex differences observed during and following prolonged cycling matched for exercise duration led to the conclusion that females have greater durability of submaximal thresholds than their male counterparts of the same fitness level. After the 90 min of cycling in the heavy intensity domain, power output associated at GET decreased only in males by 16% whereas no significant decrease was observed in females. Similarly, the power output recorded at RCP decreased more in males (−13%) than females (−6%). The male decrement in the power output associated at RCP is in line with previous findings of Clark et al. [ 2 ] and Clark et al. [ 3 ] who reported a reduction of the critical power by 8% and 11% respectively in males after 2 h of cycling in the heavy intensity domain. However, the decrement of power output at GET was slightly greater than the 6%–10% previously reported by Stevenson et al. [ 5 ] and Hamilton et al. [ 9 ] following 120 and 150 min of moderate intensity cycling, respectively. This discrepancy, despite the longer duration of both trials, is likely explained by the difference in the intensity domain in which the participants performed the fatiguing cycling task, with the present study employing cycling in the heavy intensity domain. The higher intensity sustained by our participants could have led to a greater accumulation of fatigue throughout the cycling exercise [ 35 , 48 ], as suggested by the greater magnitude of change in V̇O 2 , EE and RER, leading to a greater deterioration of performance during the ramp test performed in fatigued condition. This is not surprising, as recent scientific evidence [ 49 , 50 , 51 ] reported that the intensity at which the fatiguing cycling task is performed is more crucial in determining the subsequent downward shift in the power duration relationship in professional cyclists rather than the total duration and/or accumulated work of the trial. Although sex differences were observed at submaximal thresholds, values recorded at peak exercise, including P max (−10% ± 8%), V̇O 2peak (−7% ± 7%), V̇CO 2peak (−15% ± 12%), and BLa peak (−45% ± 20%) decreased similarly in males and females. These decreases in peak values are similar to those reported by Brownstein et al. [ 35 ], who found a 6% reduction in V̇O 2peak and a 7% decrease in P max after 90 min of cycling at 110% GET. Although the mechanism of P max and V̇O 2peak decrease after a cycling exercise in the heavy intensity domain are not fully understood, several factors could have contributed. For example, participants could have experienced substantial impairments to excitation‐contraction coupling [ 52 ], reducing their capability to express maximal power production [ 3 , 48 ]. Indeed, the modest accumulation of metabolites that occurred during the fatiguing task could have contributed to a quicker attainment of critical levels of contractile dysfunction during the final stages of the incremental test [ 53 ]. Furthermore, the reduced V̇O 2peak could be attributed to impairments in the O 2 delivery to the working muscles associated with a decline in stroke volume (and consequently in cardiac output), as evidenced by the upward drift in HR during the 90‐min cycling, likely caused by an increase in core temperature, dehydration and reduced plasma volume [ 54 ]. The greater reduction in submaximal markers of performance for males compared to females, without accompanying sex differences at peak exercise capacity, reveals insight into the potential mechanisms of the sex difference in durability. When these data are taken into consideration with those collected during the 90‐min fatiguing task (i.e., O 2 uptake and extraction, substrate metabolism, and neuromuscular function) it is possible to gain insight into the etiology of the observed sex differences. In line with previous findings [ 2 , 3 , 4 , 5 , 9 ], the 90 min of heavy intensity cycling trial resulted in a progressive deterioration of physiological parameters over time. Indeed, HR, V̇O 2 , V̇E, EE increased, whereas GE and RER decreased as exercise proceeded, indicating that the accumulation of fatigue and physiological perturbation, due to both neuromuscular and metabolic changes, resulted in a loss of movement efficiency and impairment in performance. However, as reported above, females experienced a lower decline in submaximal thresholds, which is likely related to the lower deterioration of certain physiological parameters recorded during the 90‐min task. At the beginning of the fatiguing exercise, the relative intensity of exercise (expressed as %HR peak and %V̇O 2peakPRE ) was similar between sexes (81% HR peak and 73% V̇O 2peakPRE ) indicating similar initial physiological demand. However, during the 90 min of fatiguing exercise, sex differences in the temporal change of both these two parameters appeared. Indeed, heart rate increased more in males than females from 60 min, as well as %V̇O 2peakPRE , which increased more in males than females from 15 min onwards. The different trajectories of these parameters between sexes highlight the presence of a greater V̇O 2 drift in males (4.30 mL O 2  kg min −1 ) compared to females (2.51 mL O 2  kg min −1 ), which reached 82% V̇O 2peakPRE in males and only 76% V̇O 2peakPRE in females at the end of the trial (Figure  3D ). It should however be noted that the temporal changes in V̇O 2peak over the 90 min task are unknown, and therefore data normalized to pre‐exercise peak values (i.e., Figure  4D ) likely underestimate fractional utilization of V̇O 2peak . Indeed, when V̇O 2 at the final timepoint within the 90 min task was normalized to post‐exercise V̇O 2peak , females achieved 83% ± 9% V̇O 2peakPOST , whereas males achieved 90% ± 9% V̇O 2peakPOST . This greater rise in V̇O 2 was also accompanied by a larger increase in EE as well as in O 2 extraction at the muscle level in males, as shown by their greater rise in HHbMb (+49%) than in females (+13%) as exercise proceeded. To date, the mechanisms behind the V̇O 2 drift that occurs during prolonged exercises are not completely understood [ 55 ]. However, it is reported to represent a progressive loss of skeletal muscle efficiency and homeostasis and therefore to be associated with the fatigue process [ 11 ]. Indeed, this phenomenon is documented to be associated with the fatigue‐dependent increase ATP demand of the initially recruited motor units as well as the progressive recruitment of additional (type II) muscle fibers [ 56 , 57 ], leading to a lower efficiency and a greater O 2 cost (and thus V̇O 2 ) of force production [ 58 ]. It is also reported in literature that the V̇O 2 drift is greater in those individuals that possess a higher proportion of type II versus type I fibers in the vastus lateralis [ 59 , 60 , 61 ]. This evidence supports our findings, as females are well known to have a lower proportion and cross‐sectional area of type II fibers compared to males [ 62 ]. Moreover, the greater contribution of glycolysis to ATP resynthesis and the increased metabolic instability withing the working muscles, as indicated by higher BLa values and increased HHbMb levels in males could have contributed to a greater V̇O 2 drift in males. It should be noted, however, that our data contradict previous findings whereby no sex difference in V̇O 2 drift or HHbMb increase were observed during heavy intensity cycling [ 63 ], although the duration of exercise was far shorter (30 min) compared to the present study's 90 min. Overall, our findings (i.e., the greater V̇O 2 drift, together with a larger increase in HHbMb in males), suggests that muscle contraction becomes less efficient in males, therefore requiring a higher energetic demand in order to maintain the same external power output. This greater loss of efficiency in males during the 90 min of heavy intensity cycling is a likely a contributing factor to the greater decreases in submaximal thresholds measured in the post‐exercise incremental test. With no sex difference observed for the change in V̇O 2 at GET or RCP, this implies that the change in metabolic rate at which the threshold occurred at was not different between sexes. However, the greater loss of efficiency in males meant that the metabolic rates at GET and RCP were attained at lower relative power outputs compared to females. These findings mirror those of Clark et al. [ 3 , 4 ], who observed similar changes (i.e., a higher V̇O 2 per watt of power output) with critical power following 2 h of cycling in the heavy intensity domain; however, the present study extends this to females, and demonstrates that the loss of efficiency (and corresponding decrease in submaximal threshold power) is less severe compared to males. Sex differences in the loss of efficiency could also be related to shifts in substrate metabolism during the 90‐min task. RER decreased more in females than males from 60 min of exercise, and at 45 min of exercise, carbohydrate oxidation dropped in females but not in males. In this context, the lesser decrease in RER, together with a maintained carbohydrate oxidation in males compared to females, is likely a result of the higher relative energy demand (i.e., % V̇O 2peakPRE ) as exercise proceeded, in order to maintain the target power output. Moreover, although we couldn't measure it directly with biopsies, these findings also suggests that females relied more on fatty acid utilization rather than glycogen to provide energy to the exercising muscles in the latter stages, which could elicit a glycogen‐sparing effect. A lower reliance on carbohydrate oxidation in females is supported by evidence showing that males use 25% more muscle glycogen and have higher RER values than females at matched exercise intensities below CP [ 64 , 65 ]. Indeed, as suggested by Roepstorff et al. [ 19 ], the specific muscle morphology of females characterized by a higher proportion of type I muscle fibers and a greater capillarisation in the knee extensors could explain the higher fat oxidation and the improved muscle cellular energy balance (confirmed by the lack of AMP‐activated protein kinase activation in females during a fatiguing exercise at 60% V̇O 2peak ) exhibited by females during prolonged endurance exercise. This could potentially lead to a better preservation of glycogen stores overtime and preserve performance, as glycogen depletion is an important limiting factor of prolonged exercise performance [ 52 , 66 , 67 ]. In this context, it must be acknowledged that, as reported in Figure  3F , males accumulated a greater total work compared to females (17.56 vs. 14.08 kJ kg −1 , respectively). Therefore, it is plausible that the higher relative energy demand and glycogen depletion in males are in part due to this phenomenon. Moreover, to further support this sex difference in metabolism, in our study males showed greater blood lactate concentrations at each time point of the 90 min of cycling in the heavy intensity domain. However, blood lactate did not increase progressively through the task in either group, indicating that exercise was performed in a steady state in the heavy intensity domain. The higher blood lactate concentration that was found in males than females could be explained by sex differences in muscle metabolism [ 68 ]. Indeed, it is well reported in the literature that during endurance exercise males rely more on carbohydrate oxidation to support fuel requirements and they display a greater activation of the glycolytic metabolic pathway compared to females, which is probably related to sex differences in fiber type and anaerobic metabolic properties of skeletal muscle [ 69 ]. This greater glycogen preservation in females could conceivably contribute to the lesser reductions in the power outputs at metabolic thresholds that were observed after the 90‐min task. Indeed, in states of glycogen depletion, V̇O 2 per unit of power output is increased due to the lesser ATP yield per oxygen consumed during fat metabolism [ 70 ]. Therefore, in addition to oxidative differences between sexes, differences in substrate metabolism likely contributed to the observed sex difference in durability. The decrement in voluntary force (−16% in MVIC) after the 90 min of cycling in the heavy intensity domain was a result of both central and peripheral adjustments. However, both decrements in voluntary activation as well as in resting potentiated twitches were reported without sex differences. Similarly, RPE increased in both sexes without differences. The lack of sex differences in neuromuscular fatigue contradicts previous literature showing lower levels of fatigability in females compared to males during various cycling tasks [ 15 , 16 , 17 ]. The reasons behind this discrepancy are unclear. However, as reported by Hunter [ 14 ], while there is a general consensus that females are less fatigable than males during and after intermittent and sustained submaximal isometric contractions, data concerning whole‐body dynamic contractions are more equivocal, and dependent on the task performed. Previously, it has been suggested that neuromuscular fatigue could contribute to the reduction in the oxygen cost of exercise [ 71 ] and physiological resilience [ 7 , 72 , 73 ]. Our data does not necessarily refute this, but rather it suggests that sex differences in durability are likely not due to sex differences in neuromuscular fatigue resistance. Instead, as mentioned above, sex differences in oxidative function and substrate metabolism likely play a larger role. This study is not without limitations. First, although we standardized the protocol based on the same target intensity (110% GET) and duration (90 min), the total accumulated work was higher in males than females (17.56 ± 2.82 vs. 14.08 ± 2.25 kJ kg −1 respectively, p < 0.001). As both methodological approaches can be used to compare durability between groups [ 74 ] and both have pros and cons, we preferred using a standardized duration and intensity protocol because it allowed a more controlled way to assess the changes in physiological variables at equivalent time points to better understand sex differences in the etiology of fatigue. However, we are aware that this discrepancy in the total work accumulated could have influenced the magnitude of fatigue and confound the physiological responses during the 90 min trial. Future research could consider assessing sex differences in durability during prolonged exercise when matched for total work completed. Secondly, data were collected in three different sites. Although it is well known that multicentre studies have multiple advantages (i.e., recruiting sufficiently large, and suitably powered, data set in well‐trained athletes, developing collaborations between research groups, and increase generalizability etc.), they come with potential methodological and statistical challenges. In this regard, to minimize variability in outputs and calibration and not compromise the validity of the study, we made sure to use most of the same equipment (except for the load cell and EMG system) with appropriate calibration and uniform procedures across sites. In addition, all individual data analyses were performed by the same researchers to ensure comparability between sites. However, due to logistical constraints, we could not perform any specific checks for internal validity (i.e., the same participants at each site). A third limitation is that, although males and females were well‐matched for performance level (Performance level 3) [ 31 , 32 ], we are aware that having more information about their training history and characteristics (i.e., training intensity distributions) would have provided further insight into potential variability between individuals. One further consideration is that we included participants who reached task failure prior to the 90‐min target in our analyses. This decision was made as these participants still reached an equivalent physiological “end‐point.” Indeed, the % V̇O 2peakPRE that they reached at the last V̇O 2 collection (84.4% V̇O 2peakPRE and 86.8% V̇O 2peakPRE for the male participants that ended the task at 60 and 78 min respectively; and 79.2% V̇O 2peakPRE for the female participant that reached task failure at 60 min) were in line with % V̇O 2peakPRE values of the other male (82.3% ± 6.9% V̇O 2peakPRE ) and female (75.6% ± 13.7% V̇O 2peakPRE ) participants that finished the task at 90 min. Finally, we did not weigh participants following the 90 min of cycling in the heavy intensity domain. Therefore, we could not adjust our parameters according to body mass changes pre‐post trial. In addition, it is important to recognize that the neuromuscular measures were performed 60 s after participants finished the 90‐min trial, which was standardized across testing sessions and sites; however, it could have permitted some recovery. Finally, the females in the present study were tested in a low endogenous hormonal state. Although the role of sex hormones on factors such as metabolic thresholds [ 75 ], substrate metabolism [ 76 ], and oxidative function [ 77 ] during exercise is thought to be minimal, it remains to be determined whether durability is altered by factors such as the menstrual cycle and contraceptive usage. This knowledge is crucial for female athletes and those working with them, allowing practitioners to confidently base their prescription on data generated in females.

This study demonstrated that after 90 min of cycling in the heavy intensity domain, females experienced lower reductions in the power outputs at submaximal thresholds than males, confirming our hypothesis of superior durability in females. In contrast, maximal exercise capacity was impaired to a similar degree in both sexes. No sex differences were observed in the changes in neuromuscular function, which was impaired similarly at the end of the fatiguing task. However, during the 90‐min task, females demonstrated lesser V̇O 2 drift and rise in muscle oxygen extraction, a progressively greater reliance on fat oxidation, lower blood lactate values, as well as a greater muscle oxygenation. These results suggest that the better preservation of performance exhibited by females in our study is not directly linked to sex differences in neuromuscular function. However, sex differences in substrate utilization and oxidative metabolism seem to play a more important role in delaying the loss of efficiency. The lesser reductions in submaximal thresholds in females after the prolonged fatiguing task highlight the need to consider sex‐specific training prescription and pacing strategies for long duration events. Future research should explore the role of total work and employ alternate study designs to further the understanding of sex differences in durability and how they relate to cycling performance. The novel findings of this study highlight that the deterioration of the physiological parameters that occur during prolonged cycling is different in magnitude between males and females, with the latter demonstrating a lesser derangement in the submaximal indexes of endurance performance. This could potentially have important implications for training prescription and pacing strategies during prolonged events. Indeed, the reductions in the power output associated with both GET and RCP, combined with the V̇O 2 drift, mean that in a fatigued condition maintaining the same power output might result in a rider shifting into a higher intensity domain (i.e., moderate to heavy intensity, or heavy to severe intensity), with subsequent negative effects on performance [ 78 ]. However, as shown in our study, females experienced lesser reductions in submaximal thresholds and less of a V̇O 2 slow component than males, meaning that in long duration training or competitions, females would experience a lesser likelihood of “domain drift,” or experience this later in the race/training. Therefore, athletes and the practitioners that support cyclists in their preparation for endurance events should look to base their training prescription on sex‐specific evidence.

The power that athletes can sustain during endurance events is primarily determined by the maximal O 2 uptake (V̇O 2max ), the fractional utilization of V̇O 2max at lactate threshold, and the economy during sub‐maximal exercise [ 1 ]. These variables are typically assessed in fresh conditions, however, recent studies lab‐based studies [ 2 , 3 , 4 , 5 ] have shown that these physiological pillars of endurance performance are not fixed but they progressively deteriorate as exercise proceeds. The ability of an athlete to mitigate this decrease in metabolic threshold(s) has recently been termed “resilience” or “durability” [ 6 , 7 ] and it has been proposed as a new marker of elite performance, given its important performance implications [ 8 ]. In this regard, a recent study by Hamilton et al. [ 9 ] showed that durability of the moderate‐to‐heavy‐intensity transition (VT 1 ) is an important performance parameter, as more durable athletes exhibited smaller reductions in 5‐min time trial performance following 150 min of moderate intensity cycling. In addition to lab‐based studies, recent field‐based evidence [ 8 , 10 ] suggests that the capability of a rider to maintain power following previous accumulated work might be a characteristic of successful competitive male road cyclists. Indeed, both studies reported that the more durable athletes (the ones that experienced less decline in the maximal mean power produced during maximal efforts of 1, 5, and 20 min after accumulated work) were also the higher‐level cyclists. Overall, this recent evidence highlights the relevance of competitive cyclists assessing their performance both in fresh state and after fatiguing cycling tasks, in order to quantify their “durability” and ultimately their performance. Currently, the physiological determinants of durability are not completely understood. However, the ability to reduce the magnitude of performance fatigability (i.e., decrements in neuromuscular function) during prolonged exercise is likely important in preventing the loss of efficiency (i.e., the rise in energy cost for a given work rate) that occurs over time [ 11 ]. This loss of efficiency is typically attributed to changes in motor unit recruitment profiles (i.e., greater recruitment of type II fibers owing to impaired contractile function in already recruited fibers) and substrate utilization (i.e., muscle glycogen depletion, greater fat oxidation) [ 12 , 13 ]. In this regard, it is well‐known that females are more fatigue‐resistant than males during various tasks when assessed with measures of neuromuscular function [ 14 ]. Indeed, both after 2 h of constant‐load cycling at the first ventilatory threshold [ 15 ] and following constant‐load cycling in the heavy and severe intensity domains [ 16 , 17 ] females experienced a greater preservation of contractile function of the knee extensors than males. These findings suggest that the greater fat oxidation and better preservation of glycogen stores [ 18 ], higher proportion of type I fibers [ 19 ], and greater muscle perfusion [ 20 , 21 ], as well as the superior muscle oxidative capacity of females [ 22 ], might contribute to reduce the magnitude of performance fatigability during prolonged endurance events [ 23 , 24 ]. What remains unexplored is whether these physiological sex differences promote superior durability in females, allowing them to better maintain performance compared to males during prolonged events. The importance of this is underscored by the rapid increase in the participation of females in cycling [ 25 ], as well as in the recognition of female professional cyclists with the introduction of Women's World Tour Series in 2016 [ 26 ]. Accordingly, there is an urgent need to increase knowledge through performance‐based research in females and consequently develop more sex‐specific training guidelines [ 27 , 28 ]. To date, most studies have investigated durability in male athletes and only one recent field based‐study included also female cyclists [ 29 ], reporting similar relative power profile reductions after > 10 kJ/kg, but higher relative decay in females after 20 kJ/kg for 1‐min, 5‐min, and 20‐min efforts. However, the authors acknowledged differences in training history, as well as in the physiological demands of training and competition between their cohorts of males and females, which could confound their findings. This study aimed to investigate durability and its physiological underpinnings in well‐trained male and female cyclists following 90 min of constant‐load cycling in the heavy intensity domain. It was hypothesized that (i) females would have better durability than males as measured by the decrease in maximal and submaximal markers of performance (power at GET and at respiratory compensation point, maximal power output and V̇O 2peak ); (ii) females would experience a slower derangement in cardiopulmonary, metabolic, and neuromuscular parameters during and after the 90 min of cycling exercise in the heavy intensity domain.

The authors declare no conflicts of interest.

Table S1: Descriptive statistics for the absolute values of cardiopulmonary and neuromuscular data measured in an unfatigued and fatigued state.