Rowing Performance After Dehydration: An Effect of Method

Abstract

Purpose: To investigate whether mild hypohydration in lightweight rowers compromises rowing performance despite a two-hour rehydration window. Methods: Experienced varsity rowers [11 male (82.3 ± 26.2 kg, age = 21.3 ± 4.0 years, height = 184.7 ± 2.4 cm) and three female (62.1 ± 11.8 kg, 2.3 ± 4.2 years, 166.4 ± 16.2 cm)] performed a 2000 m rowing ergometer time trial and visuomotor battery twice: once euhydrated and once after mild dehydration. Weight loss (−1.68 ± 0.23% body mass reduction) was achieved through a combination of 12 h (abstinence) of fluid and food restriction and sauna exposure. Results: Participants were significantly slower on the 2000 m rowing trial in the hypohydration condition than in the euhydration condition (+2.44 ± 4.5 s, p < 0.05). Hierarchical linear regression analyses revealed that this rowing performance decrement was explained by hypohydration achieved overnight through fluid abstinence (r² = 0.504, p < 0.01) but not by hypohydration achieved in the sauna (r² = 0.025, n.s.), corroborating our previous finding. This analysis also revealed a relationship between hypohydration-related rowing performance decrements and hypohydration-related changes in visuomotor function (r² = 0.310, p < 0.01). Conclusions: These findings suggest that rowing time trial performance can be negatively affected by relatively small changes in hydration status and that the method by which hypohydration is achieved is important. Rowing performance losses were explained by hypohydration due to prolonged fluid abstinence and by hypohydration-related changes to neural control of movement. Performance losses were not related to rapid sauna-based fluid loss.

1. Introduction

Lightweight is a division of rowing in which participants must weigh below a certain criterion to qualify for competition, providing a level of competition for smaller athletes and improving diversity in the sport [1]. Height and weight are still strongly positively related to performance in this division; so, lightweight rowers commonly weigh and train above the weight criterion and use pre-race weight loss techniques 24 h before competition to achieve the weight criterion. Acute dehydration is a prevalent technique used as it allows for weight loss without diminishing muscle protein mass [2]. Lightweight rowers must weigh in two hours before international competition, leaving a window for rehydration before competition. The degree to which acute mild hypohydration, the state of having lower total body water, negatively affects rowing performance is a subject of debate [1].

The level of hypohydration is most often measured as the percentage of weight lost as a function of the individual’s euhydrated weight, termed percent body mass dehydration (%BMD) [3,4,5,6]. This measure of hypohydration has been found to be more closely related to changes in performance than other measures, such as urine or plasma volume [7]. The effect of dehydration on performance depends both on the extent of hypohydration achieved and the nature of the performance task.

Studies on cyclists and runners suggest that participants who perform aerobically demanding tasks at 2%BMD experience performance impairments or greater in comparison to their euhydrated performance [8,9,10,11,12]. By contrast, in studies focusing on sprinting or performing maximal voluntary contractions, i.e., tasks that rely more heavily on anaerobic energy systems, performance does not seem as vulnerable to dehydration [13,14,15]. The effect of dehydration on performance may depend on the energy systems used.

Aerobic exercise may be particularly vulnerable to hypohydration [16]. Dehydration decreases blood volume, resulting in reduced cutaneous blood flow and sweating rates, and increases the core temperature [17,18,19] and fatigue [20]. Dehydration-related reductions in plasma volume trigger changes in vasoconstriction through alterations in both plasma osmolality and blood pressure. Increases in plasma osmolality are sensed centrally by subregions of the hypothalamus, leading to the release of the hormone arginine vasopressin [16]. In contrast, decreases in blood pressure are detected peripherally by baroreceptors, which stimulate increased sympathetic nervous system activity and the renin–angiotensin–aldosterone system (RAAS), triggering the release of angiotensin II [16,21].

Reductions in blood volume induce compensatory increases in heart rate and peripheral vasoconstriction to maintain blood pressure and cardiac output [22], although cardiac output does decline with hydration level [21]. As a result, during exercise, the dehydrated individual will experience increases in vascular resistance, reduced muscle blood flow, and reduced efficiency of the heart at lower levels of oxygen consumption. The performance of tasks with high aerobic demands is more vulnerable to hypohydration than anaerobic tasks due to their reliance on the cardiovascular system to both supply energy and shed heat.

According to Shepherd [23], 2000 m racing sprints are composed of 70% aerobic and 30% anaerobic activity. Although rowing contains a significant anaerobic component, studies testing rowers at 4–5.2%BMD have consistently found dehydration-related declines in rowing performance [3,5,6,24,25]. This decline occurs despite the two-hour rehydration window; so, athletes in these studies may be more hydrated than stated due to the opportunity to rehydrate before exercise. This situation leaves open the question of whether rowers’ performance is impaired at lower, more achievable levels of short-term water loss.

Whereas the rowing ergometer task is a reliable measure of power and speed in a rower, another aspect of on-water racing is that performance depends critically on oar placement and control, a facet of the task that engages mechanisms of sensorimotor control in the brain. Brain function can be degraded by hypohydration [26], leaving open another possible source of performance degradation in rowers.

The goal of the present study was to investigate whether 2000 m rowing performance is susceptible to mild hypohydration. Collegiate rowers performed two maximal-intensity 2000 m rowing ergometer time trials, once euhydrated and once following a dehydration-with-rehydration protocol that parallels the behavior of athletes on race day. We also sought to test the extent to which hypohydration-related changes in performance could be explained by the vulnerability of rowers’ visuomotor skills to hypohydration. Participants performed three tasks that measured different aspects of visuomotor control: visual motion tracking, visually controlled reaching, and response inhibition. Due to rowing’s dependence on the aerobic system and the clear susceptibility of aerobic performance to dehydration and hypohydration, we hypothesized that rowing performance would be susceptible to hypohydration. We predicted that the performance of these tasks would suffer with hypohydration and that the changes in performance on the visuomotor tasks would be related to the changes in rowing performance.

2. Materials and Methods

2.1. Participants

A sample of eleven male (mean ± 95% confidence interval for weight = 82.3 ± 26.2 kg, age = 21.3 ± 4.0 years, height = 184.7 ± 2.4 cm) and three female (62.1 ± 11.8 kg, 20.3 ± 4.2 years, 166.4 ± 16.2 cm) competitive varsity-team rowers participated in this study. We restricted participation to individuals ≥ 18 years of age who had experience rowing at the university or college level and were free from complicating medical conditions or disability status. We did not track medication use. At the time of testing, all participants were engaged in pre-season training and planned to use the study time trials as a training benchmark, but we did not formally measure their fitness levels (e.g., VO2max or lactate thresholds). With regard to our female participants, we avoided testing them during the luteal phase of their menstrual cycle. All testing was completed in early summer in Peterborough, Ontario, Canada, and all participants provided informed written consent.

An a priori power calculation (using G*Power, v.3.1.9.7; [27]) for repeated measures design and anticipated hydration-condition effect sizes of 0.33 indicated a need for 22 participants. Although we did not achieve this number, the effect sizes we report below are greater than our estimate, ranging between 0.46 and 0.79. Post hoc power calculations with N = 14 and an effect size of 0.48 yielded a power estimate of 93.5%.

2.2. Research Design

This study employed a 2-hydration (euhydrated and hypohydrated) × 2-order (hypohydration on Day 1 and hypohydration on Day 2) mixed design. Each participant performed one time trial and three visuomotor tasks once while adequately hydrated and once following dehydration with rehydration. Each participant attended on two separate testing days occurring seven days apart, and the order in which participants experienced the dehydration condition was counter-balanced. Hypohydration (goal of 2%BMD) was achieved both through 12 h of fluid abstinence and through additional time in the sauna as needed to achieve the 2%BMD goal. Participants received two hours of rest and rehydration before continuing to the rowing time trial.

2.3. Instruments and Tasks

Hydration levels were assessed using several instruments. Weight was tracked using an electronic scale (Etekfit Scale Model No.:CF351, Etekcity Corp, Anaheim, CA, USA), precise to ±0.30 kg at 49.9 kg, ±0.40 at 99.8 kg, and ±0.50 at 149.7 kg. Participants’ oral temperature was monitored as a safety precaution using an oral thermometer (BULK720, PharmaSystems Inc., Markham, ON, Canada). Thirst, a perceptual state closely associated with hypohydration, was assessed using the Visual Analog Scale of Thirst (VAST) [28], and other symptomatology associated with hypohydration [26], which are known to overlap considerably with the symptomatology of concussion [29,30], were assessed using the Symptom Evaluation Subscale of the Sport Concussion Assessment Test 3 (SES-SCAT3) [31]. Rowing performance was assessed using the Concept II Rowing Ergometer (Concept2, Morrisville, VT, USA). The rowing ergometer is widely used as a marker of rowing performance and has been shown to be closely related to actual on-water rowing performance [32]. The participants’ task was to row 2000 m in as little time as possible. Rowing time (seconds), number of strokes, and mean power (Watts) per 100 m were recorded.

Visuomotor skills were assessed on a tablet computer (iPad2, Apple Inc., Cupertino, CA, USA) fitted with custom-developed applications (programmed in Xcode https://developer.apple.com/xcode/, accessed on 1 June 2014) used to assess visuospatial updating (double-step task), target motion perception and interception (interception task), and response inhibition (stop-signal task) [33]. The code is available at https://github.com/lianabro/TabletTasks4Matlab (accessed on 5 June 2025). The tablet tasks were performed at a dimly lit office desk with minimal distractions present. Only the double-step and stop-signal tasks are reported here.

2.3.1. Double-Step Task [34]

The participants’ task was to move their index finger from the starting position, as quickly and accurately as possible, to one of three grey targets upon their appearance (Figure 1). In 75% of the trials, the target location initially shown did not change (no-jump trials). In the other 25% trials, the target location unexpectedly shifted either up or down by one position (jump trials). The jump trials were randomly interspersed with the no-jump trials, and the participants were not informed that the experimental blocks would contain jump trials. Movement adjustment was gauged through movement time (in milliseconds; ms) and end-point accuracy (in mm).

2.3.2. Stop-Signal Task [35]

Participants initiated each trial by placing their fingertip on a red start circle that appeared at the bottom of the screen. After a short and variable delay, a filled colored circle (blue or green) appeared in the center of the screen. Blue instructed participants to rapidly touch the screen anywhere to the left of the start circle, whereas green indicated a movement to the right. In 50% of trials (“go” trials), the trial ended when this movement was made. In the other 50% of trials (“stop” trials), a brief (50 ms) audible tone was presented at a particular time after target appearance. This tone indicated that no movement should be made and that the finger should remain in contact with the start circle. The relative timing of the target (“go”) signal and the tone (“stop”) signal could be one of four pseudorandomly presented stimulus-onset asynchronies (SOA): 50, 100, 150, or 200 ms. The main dependent measure was the error rate, i.e., the proportion of trials in which the finger failed to remain in continuous contact with the start circle in response to the tone.

2.3.3. Determination of Baseline Nude Body Mass

Baseline nude body mass was computed from five pretest weigh-in sessions in addition to weight on the euhydrated testing day. Participants were weighed in the morning prior to eating but after urinating and defecating. On the first weigh-in date, participants were weighed completely nude, and their own designated weigh-in clothing was measured in grams on a gram lab scale. Participants wore this clothing on subsequent weigh-ins, and nude body mass was determined from these measures by subtracting clothing weight. Participants’ oral temperature and scores on the VAST and SES questionnaires were also recorded at each session to develop a baseline score. Any single measure greater than three standard deviations from a participant’s own mean—as computed from the other four baseline-measurement days—was excluded from the baseline calculations.

2.4. Procedure

Figure 2 presents the study event timeline. Baseline weigh-ins and information sessions were conducted at the Peterborough Rowing Club. Rowing and visuomotor testing was performed at the Trent University Athletic Centre at a temperature of 20 °C on Day 1 and a temperature of 22 °C on Day 2. The facility’s saunas (>40 °C) were used to facilitate pre-weigh-in hypohydration and remained at a consistent temperature across testing days. Participants were provided with diet logs and identical training programs over the course of this study.

Participants were provided with instructions to follow the day before each rowing test day. Participants were asked to restrict their training to either rest or light steady-state exercise. Prior to euhydration testing, participants were instructed to consume fluids and food as usual. Prior to hypohydration testing, participants were instructed to abstain from fluid and food intake for 12 h prior to arrival for testing. Participants reported that they complied with instructions to abstain from fluid and food consumption over these 12 h.

On test days, participants arrived 2 h prior to their official weigh-in time. Upon arrival, participants weighed in, and their oral temperature, VAST, and SES scores were recorded. Participants in the euhydrated condition proceeded to the waiting room and continued to hydrate at will. If a participant in the dehydrated condition was determined to have reached 2%BMD (98% of their mean nude body mass), they were instructed to sit in a waiting room to maintain hypohydration until their official weigh-in time (preserving the time permitted for rehydration). Participants who had not reached 2%BMD experienced sauna exposure, in 15 min intervals, until either 2%BMD or sauna exposure reached the maximum time of 60 min. Participants were then moved to the waiting room to maintain hypohydration until their weigh-in time. In this manner, the time between the onset of fluid abstinence and the time of official weigh-in was controlled across subjects. At the official weigh-in, weight, temperature, VAST score, and SES scores were again recorded.

Our goal was to design a rehydration period that accurately reflected the lightweight rower’s practice on competition days. In the two-hour time frame between the official weigh-in and the 2000 m time trial, participants completed the tablet visuomotor tasks, rested, and rehydrated. Water, sport electrolyte drinks (Kirkland, CHO 58.33 mg/mL, Sodium 0.46 mg/mL), and breakfast foods (granola bars, bagels with jams, cream cheese, and chocolate-hazelnut and peanut butter spreads) were provided for participants to drink and eat, ad libitum. While participant rehydration and caloric intake were uncontrolled to better simulate competition, participants were instructed to drink and eat to satiety (which they also confirmed verbally). Participants were permitted to begin warming up one hour prior to testing: any length or style of warm-up was allowed, provided that each person maintained consistency across test days.

Participants completed the 2000 m rowing time trial, in groups of two, on ergometers arranged so that participants were facing away from each other. Participants used their preferred ergometer damper setting and were instructed to maintain consistency across testing days. Before beginning the time trial, each participant was asked to predict whether their performance would be better, worse, or the same as their personal best. The time trial started with a 1 min warning, followed by a ready-go instruction. Following the trial, participants were asked to rate their performance and complete the VAST and SES scales.

3. Results

For measures of hypohydration (weight, perception of thirst, and symptom reports), each participant’s measure or report provided on test days was converted to values that reflected how the measure differed from pretest measurements (see Supplementary Table S1). Percent body mass dehydration (%BMD) was calculated as follows:

$$\%\mathrm{B}\mathrm{M}\mathrm{D}={\displaystyle \frac{\mathrm{Test} \mathrm{Day} \mathrm{Weight} \left(\mathrm{kg}\right)-\mathrm{mean} \mathrm{Pretest} \mathrm{Weight} \left(\mathrm{kg}\right)}{\mathrm{mean} \mathrm{Pretest} \mathrm{Weight} \left(\mathrm{kg}\right)}}\times 100$$

(1)

The perception of thirst on test days was converted to a Z score as a function of each participant’s own mean and standard deviation based on values reported on baseline weigh-ins. The symptom questionnaire produced two measures: (1) the number of symptoms reported and (2) the average rating of symptom intensity. Again, both of these values were converted to Z scores as a function of each participant’s own mean and standard deviation from values reported on the five baseline weigh-ins. Means are reported with their 95% confidence interval.

3.1. Effect of Hydration

To check our manipulation of hydration level, %BMD (Figure 3), and Z scores for thirst, the number of symptoms of hypohydration reported and mean symptom intensity (Figure 4) were submitted to planned one-tailed paired t-tests comparing hypohydration and euhydration conditions (α = 0.05). Mean %BMD was significantly greater on dehydration day (1.7 ± 0.9%) than on euhydration day (0.3 ± 0.9%) (t(13) = 5.19, p < 0.001, d = 1.39). Mean ratings of thirst (Z) were significantly greater on dehydration day (−1.4 ± 2.1) than on euhydration day (6.2 ± 6.1) (t(13) = 4.19, p = 0.001, d = 1.11) (Figure 4A). Mean ratings of hypohydration symptom intensity (Z) were greater under dehydrated (3.5 ± 4.8) than euhydrated (0.25 ± 2.3) conditions (t(13) = 2.47, p = 0.014, d = 0.66) (Figure 4B), and the number of symptoms reported (Z) also increased under dehydrated (2.6 ± 1.9) compared to euhydrated (−0.3 ± 1.9) conditions (t(13) = 2.95, p = 0.005, d = 0.79) (Figure 4C). These data consistently suggest that participants’ hydration levels were successfully manipulated.

3.2. Rowing Performance Depended on Hydration Condition

To begin our analysis, we ruled out effects of testing order by submitting our dependent measures of rowing performance [trial time (s), mean power (Watts), and total number of strokes during a 2000 m time trial] to a 2-hydration level (euhydration and hypohydration) by 2-order (hypohydration first and hypohydration second) mixed analysis of variance (ANOVA). For all measures, this analysis revealed no main effect of order (all ps > 0.11) and no interaction between hydration and order (all ps > 0.24). These analyses revealed that there was no effect of test-day order on hydration levels and ruled out the possibility that any differences between hydration conditions are due to familiarity with the test protocol or learning.

Our measures of rowing performance were submitted to planned paired one-tailed t-tests. The mean 2000 m trial time significantly increased in the hypohydration (42.6 ± 36.3 s) compared to the euhydration (418.2 ± 37.0 s) condition (t(13) = 2.03, p = 0.034, d = 0.54) (Figure 5A). Likewise, mean power was significantly reduced in the hypohydration condition (315.9 ± 145.7 W) compared to the euhydration condition (318.3 ± 155.4 W) (t(13) = 1.78, p = 0.049, d = 0.48) (Figure 5B). A reduction in the total number of strokes in the hypohydration condition approached significance (p = 0.054, d = 0.46; Figure 5C). Together, these measures suggest that hypohydration did indeed impact 2000 m time trial performance and that these changes can be characterized as a medium effect. This finding cannot be explained by an order effect or any significant interaction between testing order and hydration level.

We also collected data regarding pace (time/100 m) achieved over each 100 m of the 2000 m time trial. We normalized the time for each 100 m interval to each participant’s own mean pace and submitted these proportions to a 2-hydration by 20-interval repeated measures ANOVA. There was a significant main effect of pace interval (F(1, 13) = 14.67, p < 0.001, η² = 1.13), such that pace was significantly faster than the average pace at every measure during the first 400 m, significantly slower than average between 700 and 1500 m, and significantly faster than average during the last 100 m. This pattern was expected as the early and late portions of the race are commonly sprinted. There was no significant interaction with hydration level and pace interval (p = 0.686), indicating that hypohydration appears to impair participants evenly over the course of their time trial.

3.3. Relationships Between Hypohydration and Performance

We conducted a series of additional analyses to determine the degree to which hypohydration explained the decline in rowing performance, using hierarchical regression to capitalize on the natural variation in hydration levels in our sample. Our participants also varied with regard to the method by which overall hypohydration was achieved; some participants experienced their peak level of hypohydration through extended fluid abstinence (%BMD_Abstinence), whereas others experienced greater acute hypohydration achieved through sauna exposure (%BMD_Sauna; see Supplementary Table S1). We conducted regression analyses to determine whether these differences impacted performance. We calculated the change in 2000 m performance across hydration conditions for each participant (where a positive value indicates that the dehydrated 2000 m time was longer than the euhydrated time) and submitted these values to hierarchical linear regression with %BMD_Abstinence on the first level and %BMD_Sauna on the second level. This model revealed that abstinence dehydration uniquely accounted for a significant proportion (5.4%) of the variance in the change in performance (F(1, 12) = 12.18, p = 0.004, η² = 0.5). Partial correlations revealed that abstinence dehydration was negatively correlated with the change in 2000 m rowing time trial performance (pr_Abstinence = −0.522, p = 0.061, VIF = 1), indicating that hypohydration due to abstinence dehydration was marginally correlated with increases in 2000 m trial time. Adding sauna dehydration to the model uniquely accounted for an additional non-significant 2.5% of the variance in the change in performance (F(1, 11) = 0.59, p = 0.459, η² = 0.02). Acute dehydration achieved in the sauna was not significantly correlated with the change in rowing trial performance (pr_Sauna = 0.225, p = 0.459, VIF = 1.745). These results indicate that declines in performance may be explained by hypohydration due to abstinence dehydration but cannot be easily attributed to acute hypohydration achieved in the sauna (see Figure 6).

To further examine this unexpected result, we selected the subset of six participants who achieved the majority of hypohydration in the sauna (marked by an asterisk in Supplementary Table S1) and used a paired one-tailed t-test to compare their 2000 m time trial performance as a function of hydration level. In this select group, mean trial time on dehydration day (415 ± 72 s) was not significantly different from that on euhydration day (415 ± 68 s) (t(5) = 0.195, p = 0.428, d = 0.22). By contrast, when we selected the subset of eight participants who achieved the majority of hypohydration through abstinence, mean trial time on dehydration day (425 ± 81 s) was significantly longer than performance on euhydration day (420 ± 84 s) (t(7) = 4.45, p = 0.002, d = 1.57).

3.4. Role of Subject Factors

Subject variables, such as sex (males vs females) (t(12) = 0.395, p = 0.700, d = 0.30), weight class (heavyweights vs. lightweights) (t(12) = −0.87, p = 0.932, d = 0.01), or ability (measured by %gold-medal standard of their euhydrated 2000 m time) (t(12) = 0.375, p = 0.714, d = 0.18), did not have any effect on the change in 2000 m time trial performance. Participants were not blind to the hypohydration condition. To address the possibility that performance changes were driven by participants’ expectations, we surveyed their expectations immediately before each time trial by asking each participant whether they expected their performance to be better, the same, or worse than their personal best. Linear regression found the participants’ prediction alone explained 35.2% of the variance in the change in time trial with hypohydration (F(1, 12) = 6.52, p = 0.025). Participants’ prediction, however, was not significantly correlated with symptom ratings, thirst ratings, or hypohydration levels (%BMD) achieved using either the acute (sauna) or prolonged (abstinence) methods (all ps > 0.195), and once prolonged hypohydration (%BMD_Abstinence) was factored into the regression (on the first level, with participants’ predictions on the second level of a hierarchical regression model), participants’ predictions accounted for a non-significant 12.5% of the variance in change in 2000 m trial time with hypohydration (F(1, 11) = 3.18, p = 0.077). Finally, Pearson correlations were computed to determine the relationship between the change in 2000 m trial time and the change in measures of self-reported well-being (thirst, experience intensity, number of symptoms, and %BMD); none of these relationships were significant (all one-way ps > 0.294), indicating that change in performance was not the result of feeling unwell or thirsty.

3.5. Performance on Tests of Visuomotor Skill

Rowers’ performance depends critically on oar placement and control, a facet of the task that engages mechanisms of upper-limb sensorimotor control. Brain function can be degraded by hypohydration [26], leaving open another possible source of performance degradation in rowers; so, we sought to test the vulnerability of rowers’ visuomotor skills to hypohydration. For each of the tests below, we ran initial analyses that included order (test conducted on Day 1 and test conducted on Day 2) as a factor, with the goal of determining whether differences between Day 1 and Day 2 could be attributed to factors like learning or increasing familiarity with the test. These analyses revealed no significant effects or interactions involving order.

Hydration levels influenced participants’ ability to adjust to an unpredictable change in reaching target position (measured by the double-step task) and affected their ability to interrupt and inhibit a planned response (measured by the stop-signal task). For the double-step task, mean movement time (MT) and mean absolute end-point error along the direction of movement from jump trials only were submitted to 2-hydration (euhydrated and dehydrated) by 2-jump direction (forward and backward) repeated measures ANOVAs. There was a significant main effect of jump direction on MT, indicating that participants needed more time to adjust their movements to target jumps requiring a reversal in direction (338 ± 37 ms) than to perturbations requiring a forward jump (259 ± 40 ms) (F(1, 13) = 43.22, p < 0.001, η² = 0.72). MT was not affected by hydration levels (F(1, 13) = 1.46, p = 0.249), and there was no interaction between hydration and jump direction (F(1, 12) = 3.37, p = 0.091). When the mean absolute error was submitted to the same ANOVA, a significant interaction between hydration level and jump direction was revealed (F(1, 12) = 4.37, p = 0.041, η² = 0.22). When the target jump required a reversal in movement direction, hydration levels influenced the accuracy of the response (F(1, 13) = 5.14, p = 0.042, η² = 0.09), such that participants’ errors were significantly larger in the dehydrated (6.7 ± 6.2 mm) condition than the euhydrated (4.4 ± 5.5 mm) condition. Hydration levels did not influence error when the adjustment was in the direction of motion (F(1, 13) = 0.049, p = 0.828, η² = 0.01).

To assess how hypohydration affected participants’ ability to interrupt a planned response, we submitted the percentage of correct interruptions to a 2-hydration level (euhydrated, dehydrated) by 4-stimulus-onset asynchrony (SOA; 50, 100, 150, 200 ms) repeated-measures ANOVA. This analysis revealed a significant interaction between delay and hydration level (F(3, 13) = 3.68, p = 0.041). Planned comparisons of hydration levels at each SOA level revealed that, at the longest SOA (200 ms), participants interrupted their planned responses more successfully when euhydrated (74.5 ± 24.2%) than when dehydrated (57.9 ± 24.3%) (p = 0.019). Together, the results from these tasks converge to show that hypohydration interfered with participants’ ability to interrupt an ongoing upper-limb response.

3.6. Do Performance Changes on Visuomotor Tasks Explain Changes in Rowing Performance?

Are declines in rowing performance explained in part by hypohydration-related changes in performance on the visuomotor tasks? To answer this question, we submitted the change in 2000 m time to hierarchical regressions with %BMD_Abstinence on the first level and measures of visuomotor performance on the second (and third, if necessary) level. The results of these analyses are presented in

Table 1. Results of hierarchical regression relationship between change in rowing performance and change in performance on tests of sensorimotor control.

Predictor Variables (Model Level)	R2 Change	pr	t(df)	p	VIF
Analysis 1: Double-Step Task
%BMDAbstinence (1)	0.472	−0.883	−1.95 (11)	0.001	1.102
Change in Movement Time (2)	0.310	0.806	7.93 (10)	0.001	1.160
Change in Accuracy (3)	0.034	0.392	2.49 (9)	0.018	1.079
Analysis 2: Stop-Signal Task
%BMDAbstinence (1)	0.504	−0.728	−3.48 (11)	0.005	1.073
Change in Success Rates (2)	0.022	0.208	0.706 (10)	0.495	1.073

Note: Analysis 1 includes predictors from the double-step task only. Analysis 2 includes predictors from the stop-signal task only.

. The analysis of change in movement time on double-step performance showed that there was a significant predictive relationship between the change in double-step performance after dehydration and the change in performance in the 2000 m time trial after dehydration. Increases in double-step MT accounted for an additional 31% of the hypohydration-related increases in rowing time above and beyond the change in performance accounted for by %BMD_Abstinence.

4. Discussion

The goal of the present investigation was to determine the effect of mild hypohydration on rowing performance. The participants experienced significant performance declines in the hypohydration condition on mean 2000 m trial time and power, on their ability to adjust to a visual perturbation during reaching, and on their ability to interrupt a planned motor response, in comparison to their euhydrated performance. These changes occurred despite a two-hour rehydration window between peak hypohydration and the rowing test, implemented to simulate rowing-competition circumstances.

Participants’ overall level of hypohydration did not explain their changes in rowing trial time. Changes in trial times were significantly explained, however, by the proportion of hypohydration participants achieved through extended abstinence. Changes in trial times were not explained by the proportion of hypohydration participants achieved acutely in the sauna. Finally, changes in rowing trial times were also explained by dehydration-related changes in performance on reaching trials, demanding an adjustment to a visual perturbation. These findings indicate that the method of dehydration may be important: levels of hypohydration achieved by abstaining from fluids and food for an extended period of time significantly accounted for rowing performance declines, whereas achieving similar levels of hypohydration acutely (in the sauna) had no effect on rowing performance. These results are consistent with our previous findings [37] and may explain discrepancies between previously reported studies.

Several other studies in which rowers performed time trials have reported significant declines in performance after prolonged abstinence with rehydration. Performance impairments in the order of 1–4 s have been found consistently on 2000 m time trials when participants achieve 4%BMD and no specific weight loss technique is mandated [5,24,25]. However, when hypohydration is prolonged and maintained over longer time frames, its effects on performance increase substantially. For example, rowers who achieved 5.6%BMD the evening before weigh-in and who maintained this level of hypohydration overnight experienced a 22 s performance decrement relative to the euhydration condition [3]. Although participants in the latter investigation faced greater levels of hypohydration (~5.6%BMD) than the aforementioned studies (~4%BMD), some degree of this weight change may be attributable to the restriction of both food and fluid.

By contrast, when hypohydration is maintained over a shorter time, performance effects are reduced. Rowers who achieved 3.3%BMD by combining fluid abstinence with an exercise task immediately before the weigh-in showed a negligible change in rowing performance [4]. In line with the findings of the present study, dehydration practices that reduce the duration of hypohydration seem to protect performance. Although it is possible that these variable findings may be attributable to other factors, the time frame and technique used for dehydration may be as important to performance as the degree of hypohydration achieved.

4.1. Differences Between Abstinence and Thermal Dehydration May Be Explained by Their Duration

Blood parameters were not monitored in the present investigation; so, we offer the following as a possible explanation for differences between abstinence and thermal (sauna) dehydration. Further research will be required to test the details of this explanation. As suggested above, it is possible that abstinence and sauna dehydration are associated with different rowing performances due to the difference in time implicit in each. While both abstinence and thermal dehydration stimulated similar declines in the total amount of body water lost, abstinence achieved this result over a much longer time frame (12 h) than sauna exposure (1 h). This difference may have distinct effects on the plasma osmolality increase expected as blood water content declines with dehydration [38]. First, we suggest that the longer time frame associated with abstinence induced greater involvement of the renal system in regulating plasma osmolality [38], limiting increases in blood osmolality in rowers who relied primarily on abstinence compared to those who relied on thermal dehydration at a given %BMD [39,40]. This hypothesis is supported by research showing that a long-duration 12 h fluid restriction protocol produced 1.5%BMD and blood osmolality levels that were not different from euhydrated controls in cyclists [41]. By contrast, acute hypohydration resulted in significant increases in plasma osmolality across four different levels of %BMD, ranging from 1.1%BMD to 4.2%BMD hypohydration, in a 2 h exercise-based dehydration protocol [17].

If this hypothesis is correct and blood osmolality is better regulated with prolonged abstinence than short-term hypohydration, this difference may change the effectiveness of the rehydration period. Elevated blood osmolality drives water from the gut into the vasculature, restoring blood volume more rapidly [38], and greater blood volume at the time of competition is associated with improved performance by maintaining cardiac efficiency and heat dissipation during maximal exertion [39,42]. Indeed, research shows that treatments designed to elevate blood osmolality—the administration of sodium pills to dehydrated participants—speed up rehydration: Participants receiving sodium while rehydrating recovered 60% plasma volume in the first hour of rehydration compared to 13% with water alone [38]. Thus, individuals experiencing short-term hypohydration after thermal dehydration may be protected from dehydration-related performance declines by faster restoration of lost plasma volume during the rehydration window.

4.2. Reductions in Visuomotor Performance Partially Explain Reductions in Rowing Performance

Together with a group-level finding that accuracy on movement reversals was compromised in the hypohydration condition, we found that hypohydration-related changes in the time needed to execute speeded, targeted movements partially accounted for hypohydration-related changes in rowing performance. Measures of visuomotor ability were initially included to probe the central nervous system functions on which the execution of proper rowing technique (control of blade timing and placement) relies. Although rowing technique plays a greater role in performance on open-water races than on the rowing ergometer, visuomotor control is involved in judging reach distance and stroke length and in maintaining balance on the ergometer, all of which are factors that can affect stroke efficiency. Recent reports indicate that golfing performance, i.e., the speeded use of a tool to hit a visual target precisely [15], and changes in balance control [40] are directly related to levels of mild hypohydration. Even small changes in the neural control of targeted reaching movements may have significant cumulative effects on overall trial time. Further research will be required to test the details of this hypothesis.

4.3. Limitations

Our participants included both male and female heavyweights and lightweights, and no effect of sex or weight class was detected, suggesting no differences across these demographics, although any sex-specific inference is limited by having only three female participants. The age range and level of experience of the sample are reflective of varsity-level rowers. While Olympic rowers tend to be slightly older and more experienced, we have no reason to believe that hypohydration would affect this population any differently. Although the mindset of Olympic athletes or athletes-in-competition may be different—elite athletes may better ignore the knowledge of being dehydrated—we found no statistically significant effect of self-predicted performance after either magnitude or time of hypohydration was factored in. The results may be affected by athletes’ knowledge that the study time trials were not performed in true competition with anyone but themselves. To better model competition events, the amount of fluid or food ingested during the two-hour recovery phase was not controlled. Participants were instructed to, and verbally confirmed, having eaten and drunk to satiety during the two-hour intermediate period, with the knowledge of the impending maximum-effort 2000 m rowing trial. We did not track participants’ intake of food and water during the rehydration period, and intake variation may have affected rowing performance. Any measure of hydration by weight following the recovery phase would have been artificially elevated by the concurrent consumption of food. This study could be improved with the inclusion of measures of change to plasma or urine osmolality as a method for verifying that changes in body weight truly reflect changes in hydration. We did not track medication use, and we did not formally measure participants’ pre-study fitness levels (e.g., lactate thresholds or maximum capacity for oxygen uptake). We will address these issues in future studies.

5. Conclusions

Consistent with our previously reported findings [32], we found that mild levels of hypohydration had significant effects on rowing performance and on measures of visuomotor control. In addition, we found that prolonged abstinence explains changes to rowing performance better than short-term hypohydration achieved in the sauna. This effect of dehydration method may help explain the wide range in performance impairment (from 0 to 22 s) found in previous investigations of rowing performance and hypohydration [3,4,5,22,23], despite similar magnitudes of water loss (3.3–5.6%BMD). We also report that a proxy for neural control of speeded motor updating explained a significant proportion of the variance in rowing performance change with hypohydration. Further research is needed both to determine safety limits for these practices and to uncover the physiological mechanism that may explain the performance-related differences between short- and long-term methods of dehydration for weight loss.

• When used as a method of cutting weight before rowing performance, acute mild (<2%) dehydration did increase 2 km rowing trial time.
• We found that the change in trial time was explained by the weight lost through abstinence and by changes to speeded motor updating.
• The change in time trial performance was not related to the weight lost through thermal (sauna) dehydration.