A retrospective longitudinal observational study examining competitive male rowers across an eight-week structured training regimen revealed that significant gains in two-thousand-meter rowing ergometer performance occurred in the absence of corresponding, statistically clear shifts in routine fixed-power physiological parameters or weekly blood-based biomarkers. Competitive athletes frequently rely on a battery of laboratory assessments, including multistage submaximal exercise tests and resting hematological profiles, to track physical conditioning and evaluate training adaptations over time. However, this investigation found that while sport-specific time-trial performance advanced reliably across the cohort, group-level trajectories for heart rate, blood lactate profiles, power output at specific lactate thresholds, and common serum markers did not exhibit parallel changes. Furthermore, individual alterations in these standard physiological markers demonstrated wide confidence intervals when correlated against performance improvements, suggesting that routine monitoring measures should not serve as direct surrogates for race-distance capacity.

Preparing for competitive rowing over a standard two-thousand-meter racing distance imposes rigorous physiological demands that recruit substantial aerobic energy production alongside critical anaerobic power contributions and sustained mechanical work capacity. To evaluate conditioning status, practitioners have historically relied on standardized two-thousand-meter ergometer trials, which offer high test-retest reproducibility under controlled indoor settings. Concurrently, sports scientists frequently employ incremental fixed-power testing protocols to assess internal cardiovascular and metabolic strain at standardized external work rates, alongside weekly resting blood analyses designed to screen for overreaching, muscle damage, or variations in hydration and protein metabolism. Despite the widespread implementation of these multifaceted testing batteries, empirical evidence demonstrating whether these separate monitoring dimensions move in lockstep during a single training cycle has remained limited, prompting researchers to analyze real-world records to determine whether internal physiological shifts truly mirror competitive performance gains.

To investigate these questions, the study analyzed routine training-monitoring data gathered from sixteen elite male rowers belonging to a provincial sports program who met the criteria of being classified as national first-class athletes or above. The final analytical group was drawn from an initial pool of nineteen rowers, following the exclusion of two athletes due to training disruption from injury or illness and one rower who lacked complete two-thousand-meter testing records. The included participants presented a mean age of 23.0 ± 1.6 years, an average stature of 195.2 ± 2.7 centimeters, a mean body mass of 97.8 ± 3.3 kilograms, and an average training background of 5.2 ± 1.3 years in rowing-specific competition within the open event category. The training program, physiological testing schedule, and blood collection routines were developed and managed entirely by the coaching and medical personnel as standard operational procedures, ensuring that no exercise load or testing variable was altered for experimental investigation.

Over the monitored eight-week period, every athlete completed an identical prescribed training volume totaling 1,826 kilometers, documented through daily post-session logging of distance, modality, and predefined intensity classifications. On-water rowing formed the primary component of training volume, accounting for 1,580 kilometers, or 86.5 percent of the total distance, while indoor ergometer training contributed 246 kilometers, representing 13.5 percent of overall exposure. Training intensity was partitioned into low-intensity training, moderate-intensity training, and high-intensity training based on standard team zones. Low-intensity endurance work predominated across the training program, totaling 1,298 kilometers and representing 71.1 percent of all classified rowing distance, whereas moderate-intensity training encompassed 506 kilometers, or 27.7 percent of the volume. High-intensity training was restricted to 22 kilometers, constituting just 1.2 percent of the cumulative distance and concentrated primarily within weeks four and eight. The athletes additionally executed twenty-eight resistance-training sessions and six dedicated core-stability workouts.

Sport-specific endurance capacity was evaluated before and after the eight-week intervention using an all-out two-thousand-meter time trial conducted on a Concept2 Model D indoor rowing ergometer equipped with a standardized damper setting of five. The two-thousand-meter completion time served as the primary performance outcome and decreased from an initial mean of 370.39 ± 5.91 seconds to 368.34 ± 6.25 seconds, reflecting a statistically significant mean improvement of −2.04 seconds, with a 95 percent confidence interval spanning from −3.36 to −0.73 seconds and an associated P value of 0.005. This improvement corresponded to a paired-samples effect size of Cohen’s dz equal to −0.83, with fourteen of the sixteen athletes, or 87.5 percent, recording faster finishing times. In parallel, mean mechanical power output recorded across the test rose significantly from 441.6 ± 21.5 watts to 448.9 ± 22.9 watts, demonstrating an average cohort increase of 7.25 watts, with a 95 percent confidence interval ranging between 2.50 and 12.00 watts and a P value of 0.005.

Internal physiological strain under controlled external work demands was evaluated through an incremental fixed-power multistage protocol conducted on identical ergometers, scheduled at approximately 10:00 a.m. following a standardized warm-up and after ensuring the avoidance of strenuous exercise during the preceding twenty-four hours. The protocol comprised six consecutive four-minute stages separated by thirty seconds of rest, executed at progressive power outputs of 170, 210, 250, 290, 330, and 370 watts, with stage progression commencing once resting heart rate subsided below eighty beats per minute. Continuous heart-rate measurements collected via telemetry demonstrated progressive increases with escalating mechanical workloads during both testing periods. However, a linear mixed-effects model testing the Time-by-Stage interaction yielded an F statistic of 0.36 with five and 150 degrees of freedom and a P value of 0.874, while the marginally averaged overall Pre-to-Post heart-rate difference across all six stages was +0.39 beats per minute, spanning a 95 percent confidence interval from −3.32 to 4.09 beats per minute with a P value of 0.828.

Capillary blood lactate concentrations sampled from the earlobe between five and fifteen seconds into each interstage rest period were analyzed using an automated analyzer to characterize metabolic power profiles across the six stages. A linear mixed-effects model evaluated on natural-log transformed values showed no significant Time-by-Stage interaction, with an F statistic of 0.93 across five and 150 degrees of freedom and a P value of 0.466. Back-transformation revealed an overall Post-to-Pre geometric mean ratio across all stages of 0.877, accompanied by a 95 percent confidence interval of 0.728 to 1.057 and a P value of 0.155, indicating that group-level lactate reductions were not statistically discernible. At the final stage of 370 watts, absolute lactate measured 5.76 ± 1.11 millimoles per liter before training and 5.22 ± 1.23 millimoles per liter afterward. Post hoc estimation of the mechanical power output corresponding to a four millimole per liter lactate concentration yielded a location shift of +10.3 watts, with a 95 percent confidence interval from −6.2 to 19.3 watts and a Wilcoxon signed-rank P value of 0.171.

Resting hematological status was tracked weekly through fasting venous blood samples obtained from an antecubital vein between 06:30 and 07:00 every Monday morning following an overnight fast, processed using automated hematology equipment. Concentrations of hemoglobin and hematocrit were evaluated across weeks one through eight via linear mixed-effects models specifying athletes as random intercepts and weeks as fixed categorical effects. Hemoglobin concentrations averaged 147.9 ± 6.0 grams per liter in week one and 147.3 ± 8.1 grams per liter in week eight, showing no statistically significant overall effect of training week, with an F statistic of 0.78 across seven and 105 degrees of freedom and a P value of 0.603. Hematocrit similarly remained stable across the intervention, registering 44.0 ± 1.8 percent initially and 44.0 ± 2.5 percent at week eight, yielding an F statistic of 0.97 and a P value of 0.461. The authors emphasized that circulating concentrations of both markers remain susceptible to fluctuations in plasma volume and hydration status during rigorous endurance programs.

Resting biochemical parameters assessing metabolic turnover and muscular stress, specifically serum blood urea nitrogen and creatine kinase, were monitored from the same weekly fasting venous samples using an automated biochemistry analyzer. Blood urea nitrogen concentrations recorded a baseline mean of 6.02 ± 1.66 millimoles per liter at week one and shifted to 7.25 ± 1.96 millimoles per liter by week eight, with linear mixed-effects modeling showing no significant overall week effect, marked by an F statistic of 0.70 and a P value of 0.672. Serum creatine kinase activity began at 263.0 ± 87.0 units per liter in week one and concluded at 278.0 ± 135.8 units per liter in week eight, producing an F statistic of 0.38 and an associated P value of 0.911. Although weekly trajectories indicated upward trends toward the final two weeks of training, the overall variation across the full eight weeks remained statistically indistinguishable from baseline variance.

To evaluate whether individual variations in physiological and biochemical responses corresponded to the observed performance enhancements, the researchers performed six unadjusted Spearman rank correlation analyses comparing two-thousand-meter improvement against changes in test parameters, computing 95 percent confidence intervals using 5,000 paired bootstrap resamples. The correlation between two-thousand-meter time reduction and the integrated heart-rate area under the curve yielded a Spearman rho of 0.38, with a bootstrap confidence interval from −0.12 to 0.76 and a P value of 0.149. The reduction in blood lactate at 370 watts exhibited a rho of 0.22, with a confidence interval between −0.37 and 0.67 and a P value of 0.421. Changes in resting blood markers from week one to week eight demonstrated weak to moderate associations that all crossed zero: hemoglobin yielded a rho of 0.38 with a confidence interval of −0.19 to 0.84, hematocrit showed a rho of 0.09 with an interval of −0.48 to 0.61, creatine kinase registered a rho of 0.12 with an interval of −0.47 to 0.65, and blood urea nitrogen demonstrated a negative point estimate of −0.24 with an interval ranging from −0.64 to 0.28.

Analysis of training exposure demonstrated that weekly training volume peaked at 266 kilometers in week two, dropped to 152 kilometers in week four, remained steady at 242 kilometers during weeks five through seven, and decreased to 198 kilometers in week eight, representing an 18.2 percent volume reduction that functioned as a pre-testing taper. Furthermore, individual trajectories revealed marked heterogeneity obscured by group averages, as blood urea nitrogen rose in eleven of sixteen rowers between weeks seven and eight, hemoglobin and hematocrit declined in ten and eleven rowers respectively, and creatine kinase increased in nine individuals while decreasing in seven, confirming that stable group means often mask substantial divergence in athlete responses.

The study authors highlighted several methodological limitations inherent to the research design, including the relatively modest cohort of sixteen male athletes, which constrained the precision and statistical power of correlation estimates and precluded generalization to female competitors or other endurance sporting disciplines. Additionally, because the investigation was conducted retrospectively using routine observational data collected during real-world training, it lacked an untrained concurrent control group to separate training adaptations from general time-related confounders. Unmeasured factors such as nutrition, neuromuscular efficiency, technical rowing mechanics, and fluid regulatory status may have further influenced both biomarker concentrations and ergometer performance. Ultimately, the investigators concluded that sport-specific competitive performance must be interpreted as a multifactorial outcome that cannot be directly inferred from single physiological or biochemical markers, underscoring the necessity of integrating multiple diagnostic streams alongside individualized training context when managing athletic preparation.

Subject of Research: Longitudinal changes in rowing ergometer performance, fixed-power physiological responses, and blood biomarkers during training

Article Title: Changes in 2000-m rowing ergometer performance physiological responses and biochemical markers during an 8-week training period in male rowers: a retrospective longitudinal study

Article References: Lyu, Z., Liu, L., & Lin, Z. (2026). Changes in 2000-m rowing ergometer performance physiological responses and biochemical markers during an 8-week training period in male rowers: a retrospective longitudinal study. Frontiers in Physiology, 17, Article 2008214. https://doi.org/10.3389/fphys.2026.2008214

Image Credits: AI Generated

DOI: 10.3389/fphys.2026.2008214

Keywords: rowing ergometer, exercise physiology, training monitoring, blood lactate, heart rate response, creatine kinase, blood urea nitrogen, athletic performance, Changes, rowing, ergometer, performance