Physiological adaptations to exercise training are specific to the type of training performed. A greater training effect is measured if both training and testing involve use of the same muscles and the same mode and speed of exercise (5,6,20). It has been speculated that adaptations to exercise training might also be specific to the time of day of training (1,11,16,21). However, there has been little scientific inquiry in this area(11,16,21).
The theoretical basis for temporal specificity in training centers on the characteristic circadian (24-h) rhythms that are demonstrated in almost all physiological variables, including many responses to acute bouts of exercise(1). For example, heart rate, oxygen uptake(˙VO2), minute ventilation (˙VE), and the rating of perceived exertion during submaximal exercise are higher when exercise is performed in the afternoon (p.m.) than in the morning (a.m.)(17,18). The time of day also influencès maximal responses and abilities, such as strength and flexibility(1) and work capacity(10,13,17,19). The greater work capacity in the p.m. is generally assumed to reflect a greater anaerobic capacity. Post-exercise blood lactate, an index of anaerobic contribution, has been reported to be higher in the p.m. (17), and maximal accumulated oxygen deficit, a measure of anaerobic capacity(7,8,14,15,22), is 6% higher in the p.m. (10). It is equivocal whether the time of day affects maximal aerobic responses(3,9,10,17,18).
If there is a time of day effect on responses to exercise training, it could mean that performance at a particular time of day will be greatest if the training has occurred at that time of day. Alternatively, temporal specificity might mean that, independent of the time of day of testing, adaptations to training will be greater if the exercise training is performed at a particular time of day. Each possibility has been addressed in one previously published study (11,21).
Hill et al. (11) have investigated the possibility of a time of day effect on responses to exercise training. They had physically active college students perform cycle ergometer training for 6 wk either in the a.m. or in the p.m. Then, all participants were tested on the cycle ergometer in the a.m. and in the p.m. at the same times of day as training sessions had been scheduled. They reported that participants in the a.m.-trained group had higher post-training ventilatory anaerobic thresholds in the a.m. than in the p.m., and participants in the p.m.-trained group had higher values in the p.m. than in the a.m. Hill et al.(11) concluded that their data provided evidence of“circadian specificity” in training, meaning that the value for ventilatory anaerobic threshold at a particular time of day was influenced by the time of day of training. However, this specificity was not evidenced in other aerobic variables, such as ˙VO2max, ˙VCO2max,˙VEmax, or maximal respiratory exchange ratio (RER).
Torii et al. (21) had participants train for 4 wk in the a.m., p.m., or evening. Before and after training, all participants were tested in the p.m. at the same time of day as the p.m. training sessions had been scheduled. There was a significant increase in ˙VO2max only in the participants in the p.m.-trained group. Training at another time of day, whether a.m. or evening, did not have a significant effect on˙VO2max measured in the p.m. Thus, training resulted in an increased ˙VO2max only when testing took place at the time of day of training.
Sport scientists are often asked the question “When is the best time of day to train effectively?” The aim of the present study was to investigate the possibility that there is temporal specificity in training to increase work capacity in high-intensity exercise (i.e., to determine if thereis a “best time”). The specific purpose of this study was to test the hypothesis that time to exhaustion and oxygen deficit in high-intensity exercise at a particular time of day would be influenced by training regularly at that time of day.
Overview. The study was approved by the Institutional Review Board for the protection of human subjects at the University of North Texas. There were two experimental groups. Participants in one group (a.m.-trained group) trained only in the a.m. for 5 wk and then performed tests in both the a.m. and in the p.m. Participants in the other group (p.m.-trained group) trained only in the p.m. and then were tested in both the a.m. and p.m. The a.m. and p.m. tests were scheduled at the same time of day as the a.m. and p.m. training sessions. The a.m. and p.m. tests were performed on different days with a day of rest between tests. Three participants in each group were randomly assigned to perform the a.m. test before the p.m. test; the remaining participants performed the p.m. test first. The primary dependent variable was work capacity (time to exhaustion). In addition, resting measures(temperature, heart rate, blood pressure) and maximal exercise responses(heart rate, RER, ˙VE, ˙VO2, ˙VCO2) and anaerobic capacity (as reflected by oxygen deficit) were obtained. Data were analyzed to determine if the a.m.-p.m. differences were influenced by the time of day at which the participants had trained.
Participants. Twelve female college students volunteered to participate in the study and provided written informed consent. All of the participants were involved in recreational activities, but none of the activities included systematic training or high-intensity exercise. As part of the prestudy screening, it was verified that no participant would have to disrupt her normal sleep habits (e.g., by having to get up earlier than usual) to participate in testing. They were randomly assigned to either the a.m.-trained group or the p.m.-trained group. The six women in the a.m.-trained group were of mean (± SD) age 20 ± 1 yr, height 169± 12 cm, and weight 70.0 ± 10.9 kg; the six women in the p.m.-trained group were of mean age 21 ± 2 yr, height 167 ± 5 cm, and weight 67.7 ± 10.5 kg.
Exercise training. Participants trained 4 d·wk-1 for 5 wk. The workouts are summarized in Table 1. Training was alternately performed on the electronically braked Ergolyne 800S cycle ergometer (Mijnhart, The Netherlands) that was used for testing and on a MedGraphics CPE 2000 (St. Paul, MN). All participants performed the same number of workouts on each ergometer.
The training program was designed by the investigators to improve performance in an exhaustive test at a constant power output of 2.6 W·kg-1. Based on the results of pilot testing, the power output of 2.6 W·kg-1 was selected for the test with the intent that it would result in exhaustion after approximately 4 to 5 min. Like the testing intensity, training intensity was individualized according to body weight. The Monday workout was very specific to the test and was designed to increase participant's glycolytic power and capacity. The highest intensity workouts, performed on Tuesdays and Thursdays, were designed to develop glycolytic power and the capacity for ATP regeneration from creatine phosphate stores. The Wednesday workout was designed to increase glycolytic capacity, specifically with respect to the ability to tolerate a build-up of blood lactate.
Except as noted for the 20th workout (below), the workouts were scheduled as described in Table 1. Each training session was supervised by two investigators. These individuals provided strong verbal encouragement to each participant to complete each workout as assigned. However, exercise training intensities were reduced when necessary (e.g., early in the training period) to ensure that participants could complete the required durations. Missed workouts were carried out on Fridays, and all participants completed a total of 20 exercise sessions. No other training was permitted. Each participant verified that she had not performed high intensity exercise outside of the training sessions.
The 20th workout was replaced by an exercise test performed using the Ergolyne 800S ergometer. The results of this test were not used in subsequent analyses. The test was performed so that the results of the a.m. and p.m. tests that followed would not be confounded by a learning effect(9,10).
Data collection. Participants were instructed to get “a good night's sleep” before testing and to abstain from alcohol for 24 h before testing and from caffeine-containing beverages on the day of testing. Participants did not eat before a.m. tests, and they were instructed to avoid heavy meals before p.m. testing. Adherence to these instructions was verified before each test session. None of the participants was a smoker.
Participants reported to the Exercise Physiology Laboratory between 0700 h and 0800 h for the a.m. sessions and between 1500 h and 1600 h for the p.m. sessions. These were the same times at which the groups had been training. Oral temperature was measured using a YSI telethermometer (Yellow Springs, OH) with an attached probe that was placed under the tongue. Temperature was recorded after the participants had been seated in a comfortable chair in a climate-controlled room for 5 min. Temperature was recorded to the nearest 0.1°C. At the end of this 5-min rest, heart rate was obtained by palpation of the radial artery until two successive 15-s counts were identical. Therefore resting heart rate values were measured with a precision of only 4 beats·min-1. After heart rate was determined, blood pressure was obtained manually by auscultation.
The seat height and position of the handlebars favored by each of the subjects were noted during the early training sessions and then used for subsequent training sessions and for all testing sessions. The subjects performed a 5-min warm up at 50 W followed by a 5-min rest, seated on the ergometer. Then, the subjects were instructed to begin pedaling and to accelerate to 80 rev·min-1. When this cadence was reached, the calculated work rate (2.6 W·kg-1) was applied, and the stopwatch was started. The ergometer provides constant work rate independent of pedal cadence. However, as in the training sessions that had preceded the testing, the participants were instructed to attempt to maintain a cadence of approximately 80 rev·min-1 with upper and lower limits of 100 and 50 rev·min-1. Throughout the tests, participants received strong verbal encouragement, but they were not provided information about elapsed time during or after the tests. Tests were terminated when the cadence fell below 50 rev·min-1 for 3 s. Time to exhaustion was recorded to the nearest second.
During each warm up and exhaustive exercise test, heart rate was monitored using a hardwire 12-lead and a Quinton 633 electrocardiograph (Seattle, WA). Breath-by-breath metabolic responses were measured using a MedGraphics CPMax metabolic cart (St. Paul, MN), which was calibrated before and after each test according to the manufacturer's instructions. From the breath-by-breath data, 15-s averages were obtained, and from these averages, rolling 30-s averages were calculated. The highest rolling 30-s ˙VO2, ˙VCO2,˙VE, and RER served as maximal values for each test.
Oxygen deficit was determined as the difference between the estimated oxygen demand and the accumulated ˙VO2(7,8,14,15,22). Oxygen demand was calculated as the product of the time to exhaustion (in seconds), work rate (in watts), and the oxygen cost per watt (in mL·s-1·W-1). The oxygen cost per watt was derived on an individual basis from the steady state ˙VO2:work rate relationship (7,15,22) determined in this study during the warm ups. Although there have been concerns that this method might slightly underestimate the actual oxygen cost of exercise or, for other reasons, provide a value that is not a valid measure of anaerobic capacity(7), it is the commonly used method. There is no reason to believe that any error introduced by using this method would have had an influence on the interpretations of results in this study.
Data analyses. The data were analyzed using a two-factor analysis of variance (ANOVA) (time of day of the test × time of day of training). There were repeated measures across the time of day of the test(“time”), and subjects were nested in groups according to the time of day of training (“group”). When there was a significant time by group interaction effect, paired means t-tests were used to compare a.m. and p.m. means of the two groups of subjects. For all analyses, the significance level was set at 0.05. Results are reported as mean ± standard deviation.
The primary effect of interest was the time by group interaction. A significant interaction effect would suggest that the a.m.-p.m. patterns were different in the two groups because they had trained at different times of day. For the primary dependent variable, work capacity (time to exhaustion), this interaction effect was significant (P = 0.02). Results ofpost hoc tests revealed that time to exhaustion was not statistically higher in the a.m. in the a.m.-trained group (P = 0.07). However, time to exhaustion was statistically higher in the p.m. in the p.m.-trained group (P = 0.03). The a.m. and p.m. means for the two groups of subjects are presented in Table 2.
There was also a significant time by group interaction effect (P= 0.02) for oxygen deficit. Results of post hoc tests revealed that oxygen deficit (whether reported in mL or in mL·kg-1) was not statistically higher in the a.m. in the a.m.-trained group (P = 0.12 and P = 0.10, respectively). However, it was statistically higher in the p.m. in the p.m.-trained group (P = 0.01 and P = 0.01, respectively). Means are presented in Table 2.
Other responses to the exhaustive exercise tests are summarized inTable 3. Due to technical problems, exercise heart rate data were available for only four of the six women in the p.m.-trained group.˙VO2max (whether reported in L·min-1 or in mL·kg-1·min-1) and ˙VEmax were higher in the p.m. than in the a.m. However, there was no time by group interaction effect. Resting heart rate and oral body temperature were unaffected by time of day, by group, or by a time by group interaction.
Typically, in a test like the one used in the present study, time to exhaustion is significantly higher (+10%) in the p.m. than in the a.m.(9,10). The important finding in this study was that the pattern between a.m. and p.m. performance times was different in the a.m.-trained group compared with the p.m.-trained group. The time of day effect was exaggerated in the p.m.-trained group, and the effect was not present in the a.m.-trained group. For the women in the p.m.-trained group, time to exhaustion was 13% higher (P = 0.03) in the p.m., while it was not at all higher (-13%, NS) in the p.m. for the women in the a.m.-trained group.
Because testing was not carried out before the training, the conclusion that the improvements in work capacity were greater at the time of day of training required two assumptions. The first assumption was that there was, in fact, a training effect. This seems clear because the participants had not been training when they entered the study, and their ability to handle the workouts improved dramatically during the 5-wk training period. No woman could complete any workout during the first week without some downward adjustment in work rate, whereas they were all able to complete each workout in the final week. In addition, in pilot work with four participants who followed a training regimen similar to that used in the present study, time to exhaustion more than doubled from pre- to post-training tests. The second assumption was that, before training, the participants had a slightly higher work capacity in the p.m. This would be consistent with all previously published studies involving similar exercise tests(1,9,10,13,17,19). The decision not to perform testing before the training was based primarily on the intent to elicit a training effect without overburdening our participants. As it stands, the present design involved 23 visits to the laboratory, 22 of them requiring exhaustive efforts. We were able to complete data collection using one-third of our subject population in each of three long academic semesters without ever interrupting any of the training/testing periods with major holidays (e.g., Thanksgiving, spring break), midterms, or finals.
During exhaustive exercise at intensities similar to those used in this study, differences in time to exhaustion reflect differences in anaerobic capacity (2,12). Maximal accumulated oxygen deficit, a measure of anaerobic capacity, has been reported to be 6% higher in the p.m. than in the a.m. (10). In the present study, oxygen deficit was 36% (P = 0.01) higher in the p.m. in the p.m.-trained group. However, it was not at all higher (-28%, NS) in the p.m. compared with the a.m. in the a.m.-trained group.
Time to exhaustion and oxygen deficit are related measures, both in a mathematical and a physiological sense. Exercise time is used to mathematically determine oxygen deficit. Anaerobic capacity (e.g., as might be reflected by the measure of oxygen deficit) is a primary physiological determinant of time to exhaustion. In this study, these two indicators of work capacity demonstrated similar responses, as would be expected.
The values for resting heart rate and oral temperature were not expected to be affected by high-intensity training. These measures tended to be higher in the p.m. in both groups of subjects, as has been found in other studies(9,17,18). Similarly, maximal cardiorespiratory responses were not expected to be affected by the short-duration high-intensity training. Again, as with the resting measures, there was not a significant group by time interaction effect.
There were large standard deviations associated with the group means for time to exhaustion. The average coefficient of variation was over 60%. This suggested that the power output selected for the tests represented a relatively different intensity for the various participants despite the fact that it was calculated based on body weight. However, this should not affect interpretation of results because (a) a repeated measures design was used and(b) the coefficients of variation for time to exhaustion were similar in the two groups. The coefficients of variation for oxygen deficit were much smaller(≈35%), and the coefficients of variation for ˙VO2max were smaller still (≈15%).
The purpose of this study was to test the hypothesis that work capacity in high-intensity exercise at the particular time of day would be affected by training regularly at a given time of day. Clearly, based on the significant group by time interaction effects for both time to exhaustion and oxygen deficit, there was evidence of temporal specificity. Participants in the p.m.-trained group had greater work capacity in the p.m., whereas participants in the a.m.-trained group did not.
Previously, Hill et al. (11) have suggested that there is temporal specificity in training to improve anaerobic threshold. After 6 wk of training, the ventilatory anaerobic threshold was higher in the a.m. than in the p.m. in participants who had trained in the a.m., and it was higher in the p.m. than in the a.m. in participants who had trained in the p.m. In a group of controls who did not train, the threshold was the same in the a.m. and p.m. before and after the training. In another study, Torii et al. (21) found that 4 wk of endurance training resulted in an increase in ˙VO2max only when testing occurred at the same time of day as training had been scheduled. They had participants train in the a.m., p.m., or evening, and they were all tested only in the p.m. before and after the training period. ˙VO2max increased only in the p.m.-trained group. They concluded that adaptations to training were greater when aerobic training was carried out in the p.m. than at other times. However, as all testing took place in the p.m., what could really be inferred from the results of Torii et al. (21) is that˙VO2max is increased (more) when measured at the time of day of training than when measured at another time.
Atkinson and Reilly (1) have suggested that the evening is the best time to train because individuals self-select higher work rates in the evening. The results of the present study cannot provide insight into their theory because the participants were not permitted to select a work rate. Rather, they were pushed very hard to perform as much as physically possible. In addition, our participants did not perform training in the evening.
Based on the improvements in both time to exhaustion and oxygen deficit, it would seem that the improvement in work capacity or performance was a function of an improved anaerobic capacity. There are several factors, other than improved anaerobic capacity, that might explain the improved performance time. For example, training at a particular time of day might be associated with a faster aerobic response at the onset of exercise, resulting in less dependence upon anaerobic pathways and a “sparing” of glycolysis until later in the exercise test. Alternatively, the greater work capacity may be explained by improved efficiency during testing at the time of day of training or by psychological factors. In addition, subjects who trained in the a.m. might have adapted to exercise in the fasted state. The faster aerobic response at the onset of exercise would explain an improved time to exhaustion if there were no increase in oxygen deficit. However, although the rate of response may have improved more at the time of day of training, any small“savings” in anaerobic energy production at the onset of exercise would have minimal impact on performance compared with the large a.m.-p.m. differences in oxygen deficit, which would seem to explain the differences in performance. Improved efficiency, which might occur as a result of a reduction in the slow component of the ˙VO2 response(4), is also a possibility and would mean that the oxygen deficit was slightly overestimated at the time of day of training. But the fact that both groups had similar a.m.-p.m. differences in ˙VO2max argues against this. In addition, any such effect would not explain the large a.m.-p.m. effects in the calculated oxygen deficit. Psychological influences cannot be ruled out. However, no participants complained about testing at the time of day at which they had not trained, no participants reported feeling any more comfortable during testing at the time of day of testing, and no participants had to alter their sleep habits to attend testing sessions. Finally, the possibility that regular training in the fasted state might specifically improve performance in the fasted state cannot be excluded, although we are unaware of any studies that have suggested that performance in a 6-min test could be influenced so dramatically by slightly decreased muscle glycogen stores.
In summary, there was a significant effect of the time of day of training on the a.m.-p.m. pattern in work capacity in high-intensity exercise. Usually, time to exhaustion and anaerobic capacity are slightly higher in the p.m.(1,9,10,13,17,19). However, in this study, although participants in the p.m.-trained group had significantly higher values in the p.m. than in the a.m., participants in the a.m.-trained group clearly did not. These results suggest that there is temporal specificity in training to increase work capacity. Greater improvements seem to occur at the time of day at which high-intensity training is regularly performed.
Address for correspondence: David W. Hill, Ph.D., Department of Kinesiology, P.O. Box 311337, The University of North Texas, Denton, TX 76203-1337. E-mail: [email protected].
1. Atkinson, G. and T. Reilly. Circadian variation in sports performance. Sports Med.
2. Billat, V., J. C. Renoux, J. Pinoteau, B. Petit, and J.-P. Koralsztein. Times to exhaustion at 100% of velocity at˙VO2max
and modelling of the time-limit/velocity relationship in elite long-distance runners. Eur. J. Appl. Physiol.
3. Burgoon, P. W., G. J. Holland, S. F. Loy, and W. J. Vincent. A comparison of morning and evening “type” during maximum exercise. J. Appl. Sports Sci. Res.
4. Casaburi, R., T. W. Storer, I. Ben-Dovi, and K. Wasserman. Effect of endurance training on possible determinants of˙VO2
during heavy exercise. J. Appl. Physiol.
5. Clausen, J. P., K. Klausen, B. Rasmussen, and J. Trap-Jensen. Central and peripheral changes after training of the arms or legs. Am. J. Physiol.
6. Gergley, T.J., W.D. McArdle, P. DeJesus, M.M. Toner, S. Jacobowitz, and R. J. Spina. Specificity of arm training on aerobic power during swimming and running. Med. Sci. Sports Exerc.
7. Green, S. and B. Dawson. Measurement of anaerobic capacities in humans: definitions, limitations, and unsolved problems.Sports Med.
8. Hermansen, L. Anaerobic energy release. Med. Sci. Sports
9. Hill, D. W. Effect of time of day on aerobic power in exhaustive high-intensity exercise. J. Sports Med. Phys. Fitness
10. Hill, D. W., D. O. Borden, K. M. Darnaby, D. N. Hendricks, and C. M. Hill. Effect of time of day on aerobic and anaerobic responses to high-intensity exercise. Can. J. Sport Sci.
11. Hill, D. W., K. J. Cureton, and M. A. Collins. Circadian specificity in exercise training. Ergonomics
12. Hill, D. W. and A. L. Rowell. Significance of time to exhaustion during exercise at the velocity associated with ˙VO2max
.Eur. J. Appl. Physiol.
13. Hill, D. W. and J. C. Smith. Circadian rhythm in anaerobic power and capacity. Can. J. Sport Sci.
14. Karlsson, J. and B. Saltin. Lactate, ATP, and CP in working muscles during exhausting exercise in man. J. Appl. Physiol.
15. Medbø, J. I., A. Mohn, I. Tabata, R. Bahr, O. Vaage, and O. M. Sejersted. Anaerobic capacity determined by maximal accumulated O2
deficit. J. Appl. Physiol.
16. Perkins, D. R. Circadian rhythms in swimming.Swimming Technique
17. Reilly, T. and G. A. Brooks. Investigation of circadian responses in metabolic responses. Ergonomics
18. Reilly, T. and G. A. Brooks. Selective persistence of circadian rhythms in physiological responses to exercise. Chronobiol. Int.
19. Reilly, T. and S. Marshall. Circadian rhythms in power output on a swim bench. J. Swimming Res.
20. Saltin, B., K. Nazar, D. L. Costill, E. Stein, E. Jannson, B. Essen, and P. D. Gollnick. The nature of the training responses: peripheral and central adaptations to one-legged exercise. Acta Physiol. Scand.
21. Torii, J., S. Shinkai, S. Hino, Y. Kurokawa, N. Tomita, M. Hirose, S. Watanabe, S. Watanabe, and T. Watanabe. Effect of time of day on adaptive response to a 4-week aerobic exercise program. J. Sports Med. Phys. Fitness
22. Withers, R. T., W. M. Sherman, D. G. Clark, P. C. Esselbach, S. R. Nolan, M. H. Mackay, and M. Brinkman. Muscle metabolism during 30, 60, and 90 s of maximal cycling on an air-braked ergometer.Eur. J. Appl. Physiol.