The Effect of Stroke Rate and Intensity on Ergometer Rowing Efficiency

Abstract

The effect of stroke rate on metabolic efficiency in rowing is not fully understood. This thesis investigated if deviations from the preferred stroke rate (PSR) affect metabolic efficiency during submaximal ergometer rowing and if this effect depends on intensity. Additionally, a musculoskeletal modelling workflow was evaluated by comparing modelled and experimentally measured whole-body metabolic cost. Muscle-level metabolic contributions and activations were analysed to investigate the muscle group demands predicted by the model.

Pre-existing experimental and biomechanical data from twelve experienced male rowers were analysed. The rowers performed two different intensity conditions at 50% and 65% of their two kilometres maximum (2K max) time-trial power output. Among these fixed power outputs, the rowers used three different stroke rate conditions at PSR, PSR-15% and PSR+15%. The experimental metabolic cost was measured using a breathing gas analysis and was normalized by rowing ergometer power output. The modelled metabolic cost was estimated using a musculoskeletal modelling workflow and residual mechanical power was added separately to account for a part of the models’ limitations. A manually calibrated fibre type profile was used for the primary analysis and compared with a literature-based profile.

The experimental results show that ergometer output normalized metabolic cost was lower at 65% intensity compared to 50% intensity, indicating higher metabolic efficiency per Watt of ergometer output at the higher submaximal intensity. Across both intensities the results indicate that there is a shallow U-shaped curve relationship between stroke rate and the normalized metabolic cost. At the PSR, metabolic cost was lowest or close to lowest and metabolic cost increased when deviating from PSR. This effect was most apparent at the 65% intensity condition, where PSR+15% produced a statistically significant increase in normalized metabolic cost from PSR. At 50% intensity condition, deviations from PSR resulted in smaller and non-significant changes.

The residual-corrected musculoskeletal modelling workflow reproduced the intensity difference, but not the stroke rate dependent metabolic responses. The model predicted a larger decrease in normalized metabolic cost from 50% to 65% intensity than observed in experimental data. The model predicted a significant stroke rate effect consisting of a gradual increase in metabolic cost from PSR-15% to PSR+15% at both intensities. Therefore, the model did not reproduce the experimental U-shaped metabolic curve or the stronger PSR+15% metabolic penalty at 65% intensity. Residual mechanical power improved the absolute agreement but remained relatively constant across stroke rate conditions and did not explain the experimental response.

The vasti were the largest contributors to muscle metabolic power predicted by the model, and their contribution and mean activation increased with increasing stroke rate, while peak vasti activation decreased. However, neither vasti nor total summed activation squared showed a robust stroke rate effect. These patterns should only be interpreted as predictions of the model instead of actual physiological mechanisms because the model did not reproduce the complete experimental response. The literature-based fibre type profile produced considerably lower metabolic cost estimates than the calibrated profile and slightly changed the predicted stroke rate response.

Overall, the findings support metabolic efficiency as a factor contributing to preferred stroke rate selection, especially at higher intensity. However, the current modelling workflow was not yet able to reliably explain the muscle-level mechanisms underlying this preference. Participant personalized muscle properties, improved muscle recruitment estimation, and a more complete representation of trunk mechanics and whole-body physiological costs may be required.