Supplementary Materials

Calibrations: Tethered and semi-tethered tests
The accuracy of the electro-mechanical device (Swim-Spektro) was tested during dynamic pulling on land. Velocity values (vST) were calibrated against data recorded using 2D video analysis. The device pulled an object at four constant speeds (0.5, 1.0, 1.5, 2.0 m·s^-¹) while a static camera (Hero4, GoPro, San Mateo, CA, USA, sampling rate 60 Hz, pixel resolution 1280 x 720) recorded the object transition on the sagittal plane (in a calibrated space 4 m in length); based on this calibration procedure, speed data, as determined using the Swim-Spektro (vSpektro), were corrected using the following linear equation:

vST = 1.0024 · vSpektro + 0.006 (R² = 0.99).

Force values (imposed loads) were calibrated against data recorded using a (previously calibrated) load cell (Globus Ergometer, Globus™, Codognè, Italy, range 0–2500 N, sampling rate 1000 Hz): a participant was asked to run on land at maximal speed while pulling imposed constant loads (30, 60, 90 N, loadSpektro) and with the load cell (Globus) positioned in series with the cable (between the participant and the Swim-Spektro device). Based on this calibration procedure, the imposed loads (loadSpektro) were corrected using the following linear equations:

load(Globus) = 1.1831 · loadSpectro + 30.467 (R² = 0.99).

Calibrations: Passive drag measurements
Towing speed (vTOW) was calibrated as described for the semi-tethered tests (Swim-Spektro device). Towing force was calibrated by pulling constant loads (26, 40, 68 N, loadSpektro) at a constant speed (1.0 m·s^-¹); in this case:

load(Globus) = 1.0311 · loadSpektro – 2.2117 (R² = 0.99).