The Role of Anaerobic Power Reserve on Exercise Tolerance: From Physiological Correlates to Performance Prediction

Introduction The anaerobic power reserve (APR) is defined as the difference between maximal sprint power and the power associated with V̇O2max and has been proposed as a simple tool to capture the athlete’s anaerobic characteristics. However, its actual anaerobic role in prescribing severe exercise intensity and its predictive value for short-duration cycling and swimming performance remain unclear. Therefore, this thesis aimed to: (i) determine whether APR reduces the heterogeneity of physiological responses during V̇O2max-based HIIT; (ii) examine the relationship between APR and time to exhaustion across intensities; (iii) evaluate the combined predictive value of APR and critical speed for 50-, 100-, and 200-m frontcrawl performance; and (iv) explore the association between work performed above critical power, estimated using different models, and multiple anaerobic indices. Study 1. This study examined whether APR and glycolytic power reserve (GPR) could reduce variability in HIIT responses compared with maximal aerobic power (MAP) based prescription. Twelve trained cyclists completed incremental and Wingate tests to determine MAP, APR, and GPR; then performed three randomized HIIT sessions to exhaustion (60 s work:60 s active recovery) prescribed using MAP, APR, or GPR. No significant differences in variability were observed across prescription methods for exercise tolerance, physiological, or perceptual outcomes. These findings indicate that APR- and GPR-based prescriptions do not reduce heterogeneity in HIIT tolerance or responses relative to MAP, suggesting limited ability to distinguish individual aerobic–anaerobic profiles. Study 2. This study investigated associations between APR, time to exhaustion (Tlim), work above critical power (W′), and maximal accumulated oxygen deficit (MAOD) in endurance-trained male cyclists, and compared APR with a maximal power reserve (MPR) model using critical power (CP) as the lower boundary. Nineteen cyclists performed multiple trials to exhaustion at different intensities (130 to 80% of MAP) and a Wingate test. APR and MPR correlated with most Tlim, except at lower intensities, with associations remaining only for supramaximal efforts after controlling for CP or MAP. When peak power output was controlled, only MPR remained associated with Tlim. Both reserves correlated with MAOD and W′, but only MPR remained related to MAOD after adjustment. Overall, power reserves were related to exercise tolerance, particularly at high intensities, but these relationships were largely driven by peak power output rather than the choice of lower boundary. Study 3. This study examined the relationship between modified 3-minute all-out test (3mAOmod) parameters and swimming performance over 50-, 100-, and 200-m front crawl, and evaluated the predictive accuracy of critical speed (CS) and multiple linear regression (MLR) models applying the anaerobic speed reserve (ASR) framework. Twenty-three competitive youth swimmers performed the 3mAOmod and subsequently competed in official races. CS showed increasing correlations with race time as distance increased, whereas correlations for D′ and ASR decreased with distance. The CS model underestimated performance in the 50- and 200-m events but not in the 100-m race. In contrast, MLR models incorporating CS, D′, or ASR, and anthropometric variables accurately predicted performance across all distances. These findings support the physiological relevance of 3mAOmod parameters and demonstrate that multivariate models provide more accurate and practical performance predictions than the CS model alone. Study 4. This study compared the estimated curvature constant (W′) and actual work performed above critical power (W>CP) across intensities using different mathematical models and examined their relationships with anaerobic indices. Twenty-one participants completed trials to exhaustion at multiple intensities and a Wingate test. Critical power and W′ were estimated using work-time, inverse-time, and hyperbolic models. W>CP varied with both exercise intensity and model, decreasing at the highest intensities. Two-parameter models showed lower estimation error and stronger associations between W′ and anaerobic markers. Regardless of the model, W>CP at supramaximal intensities correlated with accumulated oxygen deficit, lactate, and Wingate performance. In contrast, W′ was associated with anaerobic markers only in specific models. These results indicate that anaerobic characteristics cannot be fully described by a single parameter and that model selection substantially influences interpretation. Conclusion. Overall, this thesis shows that the APR is an intuitive but physiologically incomplete construct. While APR is moderately associated with anaerobic capacity and exercise tolerance at (supra) V̇O2max intensities, its prediction accuracy is limited, intensity-dependent, and largely driven by peak power output rather than by the chosen lower boundary (MAP or CP). Consequently, APR does not improve the normalization of physiological responses or tolerance during HIIT compared with conventional MAP-based prescriptions. However, when applied to short, high-intensity efforts (such as 50-200 m front crawl or cycling effort < ¬⁓5min), APR is strongly related to performance. Therefore, these findings indicate that APR is unsuitable for standardizing HIIT intensity because it cannot capture the complexity of the dynamic depletion and reconstitution of W>CP; however, it could be used as a simple and useful tool to prescribe and predict short-duration performance.