Pre-experimental sessions
Prior to the main trials, each subject visited the laboratory four to five times for familiarization with instrumentation, methodology, and anthropometric evaluation. In addition, each subject performed two-legged submaximal (4-stages with 4 min duration), or maximal exercise on an electrically braked cycle ergometer (Lode RH, the Netherlands) to determine the oxygen cost-work rate relation and the VO_2peak. VO_2peak was determined as the highest VO₂ average of the 30 s, provided that the following criteria were fulfilled: 1) a plateau in VO₂ despite an increase in work rate, 2) respiratory exchange ratio higher than 1.1, and 3) blood lactate concentration higher than 9 mmol·L^-1. Lactate was measured (Accutrend, USA) 3 minutes after the completion of exercise

Procedure
Three days prior to the main trial, the subjects were asked to record their diet and physical activity and replicate these habits before each experiment. In addition, the subjects had to abstain from physical activity, caffeine and alcohol consumption during the day before the experiments. The experiments were conducted at the same time of day, at least 3 hours after a light meal. After the subjects emptied their bladder, their body mass was measured (Bilance Salus, Italy) and electromyography (EMG) and electrocardiography (ECG) electrodes were fixed to the knee extensors (vastus lateralis and vastus medialis) and to the chest, respectively. Also, a thermistor was placed in the rectum, at a depth of 13-15 cm from the sphincter. Following this preparation, the subjects warmed up on the cycle ergometer for 8 min at 50% of their VO_2peak. After 5 min stretch, maximal voluntary isometric knee extension (MVC) was determined for knee extensors using a modified strength-training device connected with a calibrated load cell. Three 5-sec maximal trials with strong verbal encouragement were conducted with one-minute rest, in order to obtain the maximum EMG signal. A fourth trial was conducted if there was a difference greater than 5% between trials and then the two trials with the higher force were averaged. Afterwards, subjects relaxed for about 20 min on the cycle ergometer while instrumented. Duplicate 10 µl and 50-60 µl capillary blood samples were collected from the right index finger for haemoglobin (Hb) and haematocrit (Hct) determination, respectively. In the first condition, subjects cycled at 80 rpm for 50-55 min with two legs at 60% of their 2-legged VO_2peak. The second condition was conducted at least three days later and the external workload was adjusted in order to attain the half of the VO₂ observed in the first condition. The same oxygen uptake per leg was adopted in order to avoid higher metabolic response and energy turnover demands in 1-legged condition in case that half work rate (Watt) per leg was selected (Jensen-Urstad et al., 1994). The absolute exercise intensity was 126 ± 1 Watts in 2-legged exercise, whereas in the 1-legged condition was calculated according to Jensen-Urstad et al., 1994 to be ~9% less than the half of the absolute intensity in 2-legged (work rate: 126/2-9% = 57 Watt). During the 1-legged condition, the non-exercising foot was rested comfortably between pedals on the bike frame, and the return of the pedal was achieved by elastic springs attached to hand-bar. Subjects were familiarized with the experimental setup in 1-legged exercise at least one day before the experiment for 20 min. To secure the development of CV_drift subjects cycled without fan or fluids. In all experiments, the environmental temperature was 23.5 ± 0.4 °C and the relative humidity 56.2 ± 4.1%.

CO was determined from 15-22, 30-37 and 47-54 min using the exponential CO₂ rebreathing method (Defares, 1958) with triplicate measurements separated by 2 min to ensure physiological steady state. In case those measurements differed more than 5%, a fourth maneuver was performed and all trials were averaged. VO₂ was measured breath by breath and averaged every 10 sec at 1-8 min of exercise and, thereafter, 2-3 min before each CO determination. Muscle oxygenation (oxygen saturation, StO₂) and blood volume (total hemoglobin, HbT) were measured by near-infrared spectroscopy. One minute EMG activity was stored at the 15^th, 25^th and 40^th min. Double measurements of blood pressure were taken at rest, at the ~25^th min (before CO measurement) and before the end of exercise with an automated system (Criticon Dinamap, Finland) in a standardized position with the arm supported by the researcher at the level of the heart . In case of a difference between measurements greater than 5mmHg, a third measurement was conducted. HR (Polar S- 810, Finland) and body temperatures were recorded every 5 min. Moreover, HR was recorded beat by beat in a 5 min period at rest and at 8-13 and 37-42 min of exercise. These HR data were used for HR variability (HRV) analysis. At the same periods, respiratory frequency (Rf) was measured at 500 Hz with a fast response thermocouple (TSD 202A, Biopac systems, USA) attached to the respiration valve. Forearm skin blood flow (FskBF) was recorded with Laser Doppler Flowmetry at rest and for 1 min at ~15 and 44 min of exercise. The general and local (legs) feeling of fatigue was expressed every 5 min using the Borg scale (RPE).

Analytical methods and equipment
A metabolic cart was used for the measurement of VO₂ (MedGraphics, CPX-D, USA), which was calibrated before exercise with two different gas mixtures. CO was determined with the CO₂ rebreathing method (Defares, 1958) as modified by DaSilva et al., 1985 and Nassis and Geladas, 2002. Subjects breathed from a bag containing 1.5-2 times the tidal volume at 50 breaths per minute, set by a metronome. According to CO₂ exponential method, CO was calculated using the indirect Fick equation. The three measurements of CO at a specific time period were averaged. The coefficient of variation in different exercise intensities varied from 2.4 to 3.6%, which ensures the reliable measurement of CO, given that the respective values in a direct method are 5-10% (Warburton et al., 1999).

Changes of StO₂ and HbT were recorded using near infrared spectroscopy (NIRS) (In Spectra³²⁵, Hutchinson Technology Inc, USA). The operational principles of NIRS device are given elsewhere (Kounalakis et al. 2008). The 25 mm NIRS probe was attached and secured with an adhesive patch to the shaved and cleaned skin in the medial aspect of the rectus femoris muscle. Before each testing the NIRS device was calibrated with a single light-scattering standard and validated against standard references equivalent to 38% and 90% hemoglobin saturation.

EMG activity, indicating relative muscle activation level, and MVC of knee extensors were measured in knee angle of 60° (0° = fully extended). Velcro straps were used to stabilize the subject and the mechanical response was recorded by a force cell (LC-500F, Kyowa, Japan) and by a dynamometer (SS- 25, Biopac, USA). EMG was recorded using circular electrodes (Kendall-Arbo, Germany) (bandwidth of 20-500 Hz sensitivity of 0.08 mV) placed over the vastus lateralis (VL) and medialis (VM) belly with an inter-electrode distance of 20 mm. Before electrode placement, the skin overlying the muscles was carefully shaved and cleaned with an alcohol swab. A 50-Hz line filter was applied to the EMG data to prevent interference from electrical sources. EMG and force signals were digitized online (TEL 100, BIOPAC System, USA), sampled at 1 kHz and stored for further analysis. EMG signals were high pass filtered with a cut off frequency of 0.2 Hz and then smoothed with a low pass filter with a cut off frequency of 400 Hz (Acknowledge 7.3.3 Software, Biopac USA). The same EMG analysis for 15 consecutive cycles during cycling was conducted. All EMG data were normalized by dividing the value at each time point during cycling by the averaged EMG value obtained during the MVC and expressed as a percentage.

Five min fast Fourier transformation was used for HRV analysis, which is considered to assess autonomic influences of the heart (Carter et al., 2003). An ECG100 unit (Biopac, USA) amplified the signal from electrodes and digitized it online at 1 kHz. Frequencies of 0.04-0.15 Hz (LF-ms²) and 0.15-0.40 Hz (HF-ms²) were normalized according to total power spectrum and very low frequencies (<0.04 Hz, VLF-ms²): LFn=LF/(total power-VLF)x100 and HFn= HF/(total power-VLF)x100. The ratio of LF·HF^-1 was used as an index of the degree of sympathetic activation (Carter et al., 2003). Spectral analysis of the respiratory signal was performed by the same procedure. These spectra were used to assess the main respiratory frequency and to locate the respiratory component of the power spectral analysis of respiratory rate-interval variability.

Rectal temperature (Tre) was measured with a thermistor (Yellow Springs, USA) connected to a tele-thermometer (Yellow Springs, Model 46, USA). FskBF was estimated from the microvascular blood cell velocity measured with a Laser Doppler probe (LDF 100A, BIOPAC System, USA) placed on the dorsal side of the right arm and secured to remain in the same position and in full contact with the skin for the experimental trial. Laser Doppler was appropriately calibrated before its use.

The laser Doppler probe was connected to a LDF100A flow module (BIOPAC, USA). All sensor signals collected were first elaborated by a UIM100A A/D interface (BIOPAC) which was connected to the MP100A data acquisition unit (BIOPAC) for storage and further analysis. Hb concentration was determined with the cyanmethaemoglobin method (Dr. Lange Mini-Cuvette, LKM 143, Germany). Hct values were obtained via a reader (Hawksley, UK) after a 4-min centrifugation at 11500 rpm. Coefficients of variation in repeated analyses were 1.8% and 1.7% for Hb and Hct, respectively.

Calculations
SV (ml·beat^-1) was the ratio of CO over HR. MAP was calculated from systolic (SP) and diastolic pressure (DP): MAP=DP+1/3·(SP-DP). Total peripheral resistance (TPR) was MAP·CO^-1, and arteriovenous oxygen difference (a-vO₂Diff) was VO₂·CO^-1. Changes in plasma and blood volume were calculated using Hb and Hct data (Dill and Costill, 1974).

Statistical analysis
A two-way analysis of variance with repeated measures on both factors was used (Statistica 5.0, USA). Analysis of Covariance with repeated measures was also employed for ∆SV with oxygen uptake as covariate, for HR with Tre as covariate and for HRV using the Rf as a covariate. A Tukey test was employed to assign specific differences in the analysis of variance. Data are presented as means ± SD. Backward stepwise multiple regression analysis was conducted using as dependent variable the changes in HR and, as independent variables, the Tre, EMG activity, LF HF^-1 ratio and BV and PV changes. The same procedure was conducted with changes in SV as dependent variable and ∆HR, ∆Tre, ∆BV, ∆SBF, LF HF^-1 ratio as well as ∆HbT as independent variables. Significance level was set at 0.05.

The mean VO₂ was 1.04 ± 0. 09 and 2.1 ± 0.12 L·min^-1 for 1- and 2-legged condition, respectively, with no changes over time. The intensity in 2-legged cycling was at 58% of 2-legged VO_2peak and the respective intensity for 1-legged condition was calculated according to Klausen et al., 1982 to be ~42% of 1-legged VO_2peak.

CO was higher in 2-legged condition (p < 0. 01) with a tendency to decline (0.9 L drop) at the end of exercise (p = 0.07, Figure 1a). In both conditions SV dropped after the 35^th min in 2-legged and at the 50^th min in 1-legged exercise (Figure 1b). At the 50^th min of exercise ∆SV was -20.8 ± 0.8 ml·beat^-1 in the 2-legged and -13.3 ± 1.3 ml·beat^-1 in the 1-legged exercise. HR had the reverse course of SV with differences between conditions (p < 0.05) (Figure 1c). The HR drift was still greater in 2-legged exercise when the percentage of VO₂, was used as a covariate (p = 0.04).

MAP was 101 ± 1. 3 and 100.4 ± 1 mmHg in 1-legged and 2-legged condition, respectively, without significant differences over time. TPR was higher (p < 0.05) in 1-legged exercise at the 20^th min (8.42 ± 0. 38 vs. 5.92 ± 0.27 mmHg·L·min^-1), as well as at the 50^th min (9.2 ± 0.41 vs. 6.26 ± 0.31 mmHg·L·min^-1) of cycling. In contrast, a-vO₂Diff was higher in 2-legged condition (8.7, 8.6 and 9.2 and 11.2, 11.8 and 12.4 ml of O₂ per 100 ml of blood for 1- and 2-legged conditions at 20, 35 and 50^th min, respectively, p < 0.01).

The differences between conditions in StO₂ were 7-10% (p < 0.05) from the beginning of exercise, while the respective differences in HbT did not reach statistical significance. Both variables kept increasing from the 25^th min (Figure 2).

Integrated EMG activity was higher for VM and VL in the 2-legged condition (Figure 3) due to higher external workload applied. The LF·HF^-1 ratio was increased from rest in both conditions with higher values observed in 2-legged cycling (2.98 ± 0.45, 5.23 ± 0.54 and 5.84 ± 0.58 for 1-legged vs. 3.40 ± 0. 39, 8.04 ± 0.49 and 8.22 ± 0.86 for 2-legged condition at rest, at 9-14^th and at 35-40^th min, respectively).

Percent of plasma and blood volume decline from the pre-exercise values were significant in the 2-legged condition, while differences between conditions were observed only in the first variable (Table 1).

At the end of exercise rectal temperature was higher compared to rest in both conditions (p < 0.05). At the 35^th min of exercise and afterwards, Tre was higher in the 2- legged condition (Table 1). FskBF increased throughout exercise compared to rest in both trials, but no difference was detected between conditions. The general and local RPE was different between conditions after the 30^th min of exercise. The mean values of local and general RPE for one and two leg exercise were 12 ± 0.5 vs. 13.7 ± 0.3 and 10.5 ± 0.4 vs. 11.8 ± 0.4, respectively.

In the 2-legged condition, multiple regression analysis revealed that ∆SV was influenced (r² = 0.99, p < 0.001) by ∆HR (p < 0.01), ∆BV (p = 0.04) and ∆HbT (p = 0. 02), while ∆HR was explained (r² = 0.88, p < 0.01) by ∆Tre (p < 0.01) and ∆EMG activity (p < 0.01). In 1-legged condition, ∆SV was explained (r² = 0.99, p < 0.001) by ∆HR (p < 0. 01), whereas none of the examined factors explained ∆HR.

• The magnitude of the participated muscle mass plays a critical role for the development of cardiovascular drift, when the oxygen consumption per leg is the same.
• Apart from thermal status and blood volume decline, central command plays a role on cardiovascular regulation during steady state exercise performed with large muscle mass.