Content uploaded by Ferran A. Rodríguez
Author content
All content in this area was uploaded by Ferran A. Rodríguez on Dec 11, 2018
Content may be subject to copyright.
Chapter 11
SIMULATED METABOLIC RESPONSES DURING HIGH-
INTENSITY INTERVAL TRAINING BASED ON A
MATHEMATICAL MODEL IN ELITE SWIMMERS
Philippe Hellard1, Ferran A. Rodríguez2, Carl Dupont1, David. B. Pyne3,4,
Alois Mader5, Sebastian Weber6
1Research Department, French Swimming Federation, France
2INEFC-Barcelona Sport Sciences Research Group, National Institute of Physical Education of Catalonia,
University of Barcelona, Barcelona, Spain
3Department of Physiology, Australian Institute of Sport, Canberra, Australia 4Research Institute for Sport and
Exercise, University of Canberra, Canberra, Australia
5German Sport University Cologne, Germany
6INSCYD, Switzerland.
ABSTRACT
High-intensity interval training (HIT) has been described as a training method that strongly stimulates
the mechanisms of metabolic, neuromuscular and technical adaptations. Studies focusing on the effects
of HIT in swimmers are rare despite its widespread use in practice. It is thought that HIT can
powerfully stimulate the three energy systems, both in metabolic and technical terms, to a level close to
competitive performance. In high-level swimmers, it can be difficult to manipulate all HIT load
parameters (distance, speed, rest, recovery). Computer simulation using mathematical modeling of
energy metabolism may provide an effective alternative solution to coaching trial and error. For 7 elite
male swimmers (21 ± 4 years, 77 ± 8 kg, 188 ± 9 cm) metabolic responses were simulated for maximal
interval sets (20 x 25 m, 12 x 50 m, 12 x 100 m and 6 x 200 m with short, medium or long recovery).
Mixed-effects regression analysis was used to model the association between the swimmers’ metabolic
responses and their metabolic profile (maximal oxygen uptake and lactate production rate, V
̇O2max and
V
̇Lamax, respectively), adjusted for speed, interval length and recovery interval. The aerobic metabolism
(mean V
̇O2, %V
̇O2max, ΔV
̇O2 and % aerobic contribution) was most solicited in those swimmers
exhibiting higher V
̇O2max and lower V
̇Lamax, and for long intervals and short recovery periods (p <
0.05). Glycolytic metabolism (meanV
̇La, %V
̇Lamax, ΔV
̇Lamax and % glycolytic contribution) were
greater in those swimmers with high V
̇Lamax and for fast speeds and long rest periods (p < 0.05). The
simulation of metabolic energy responses using a mathematical model is a promising method to
characterise the internal training load during HIT in elite swimmers.
* Corresponding Author Email: hellard.ph@gmail.com
Keywords: swimming, bioenergetics, energy metabolism, modeling, computer simulation
INTRODUCTION
High intensity interval training (HIT) is now considered one of the most effective training methods to
improve cardiorespiratory, metabolic and neuromuscular functions and, in turn, athletic performance [1, 2].
HIT is typically defined as short (< 45 s) to long (2-4 min) bouts of high-intensity exercise interspersed by
periods of recovery. Programming effective interval training sessions requires the balanced coupling of
several variables including the work interval intensity and duration, the recovery interval intensity and
duration, the number of repetitions, the number of series, and the intensity and duration of the between-
series recovery periods [1, 2, 3, 4, 5].
It is widely accepted that the time spent above 90% of V
̇O2max is effective in stimulating central and
peripheral cardiovascular maximum adaptations [1, 2, 4, 5]. In practice, it is recommended to use short and
long HIT intervals with a work:recovery ratio >1:1 and a minimum total time at V
̇O2max of 10 min. Work
intervals longer than 4 min and intensity equivalent to 90-95% of V
̇O2max are recommended to produce
cardiovascular adaptations. Based on scientific work it appears that HIT added to previous high training
volumes can increase both intense and prolonged endurance performance by 2-4% in well-trained athletes
[1, 2].
In addition to the time spent at intensities near V
̇O2max, there are many other parameters that determine
the stimulus induced by a HIT session, including the anaerobic contribution, neuromuscular solicitation and
technical quality of the exercise [1, 2]. From the metabolic standpoint, HIT workouts with short
intervals (20 to 30 s of work with 20-30 s of recovery) performed at the intensity near to V
̇O2max reduce the
glycolytic contribution to exercise (blood lactate <5 mmol·l-1). In contrast, all-out sets longer than 20 s with
long recovery completely mobilize the glycolytic anaerobic sources (blood lactate >10 mmol·l-1). Thus,
lengthening of the interval or HIT session duration without change of recovery duration or intensity clearly
increases the glycolytic anaerobic contribution, as more work is conducted in a given time [6].
The picture in swimming may be somewhat different to other sports. Swimming performance depends
on the ability to mobilize the most energy possible in the time allowed using the three metabolic pathways
(aerobic, glycolytic and phosphagenic), it is and is largely determined by the ability to spend the least
possible energy at race speed [7]. The energy cost reduction is associated with a superior increase in
performance relative to a similar increase of metabolic power [8, 9]. Therefore, in swimming, over a
swimmer’s career and during the course of a season, the coach will seek to increase the metabolic and
physical potential, with a focus on diminishing the energy cost. Consequently, different HIT sets should be
programmed in relation to the technical and metabolic requirements of competition events. The challenge
for the coach is prescribing training-specific sets accordingly to the contribution of aerobic (Eaer) and
anaerobic alactic (Ean,al) and lactic (Ean,lac) metabolic processes to the total energy expenditure as a function
of the swimming event (e.g., for 100 m: Eaer 41%, Ean,al 20%, Ean,lac 39%; for 400 m: Eaer 73 %, Ean,al 6 %,
Ean,lac 21%) [10].
Logically, the proportions of the various HIT training sets should mimic the metabolic intensities in
similar proportions to those induced by the competitive events. Paradoxically, the quantitative analysis of
training intensities in high-level swimmers reveals different percentage distributions to those reported for
competitive events. For example, while ~40-60% of the energy provided during a 100 m swim is anaerobic
[10, 11, 12], a maximum of 8-12% of training time is usually programmed by coaches for sprint swimmers
[13, 14]. The explanations for this low proportion of anaerobic training largely relate to an increase in
metabolic and muscle fatigue during swimming at high intensity inducing a decrease in propelling
efficiency [15, 16, 17].
In swimming, effective programming of HIT intervals (speed, distance and recovery) is achieved by
combining metabolic load with technical efficiency (i.e., swimming at high or maximum intensities while
preserving a good technical skill) [15, 18]. From this perspective, some studies have shown that short HIT
bouts enable swimmers to exercise at intensities closed to race speed and for a longer time than continuous
training [3, 19]. On the other hand, for maximum intensities equal and greater to the speed at V
̇O2max
(vV
̇O2max), short interval training may be preferred if the aim of the session is to keep a high stroking
efficiency (i.e., high SL / lower SR[1]). This outcome can be achieved while maintaining a large volume of
HIT at high speed without metabolic overload [20]. Both for coaches and scientists [21, 22, 23], frequent
programming of HIT sessions enables the swimmer to acquire an efficient technique at high stroke rate
(frequency ≥ 40 cycles·min-1), combined with the development of his/her metabolic capabilities such as the
anaerobic lactic capacity, muscle buffering and lactate transport.
During HIT sessions the metabolic energy contributions are mixed [8, 12], V
̇O2 is maximal at central
and peripheral levels [24, 25] and the remaining energy requirement is supplied by the anaerobic pathways
(i.e., ATP-PCr and glycolytic systems) [26, 27]. In relation to V
̇O2 kinetics responses and the time limit of
exercise as a function of swimming speed, Sousa et al. [24], in study with 12 national level swimmers,
reported faster slow-component V
̇O2 kinetics at 105 than 100% vV
̇O2max and an inverse relationship
between intensity and time to exhaustion. The maximum power with anaerobic lactic metabolism is
achieved during all out exercise for 15 to 30 s [26]. The total anaerobic expenditure (i.e., O2 deficit,
anaerobic capacity) increases substantially up to 2-3 min of exercise and decreases for longer periods [27,
28]. Thus, it would seem that both the ATP-PCr and glycolytic systems are maximally loaded in HIT
intervals for 1 to 3 min, contributing to the improvement of the physiological processes related to the
maximum O2 debt. Ogita [27] also compared continuous training at 70% V
̇O2max for 60 min to short-
interval training consisting of seven to eight sets of 20 s at 170% V
̇O2max with 10 s recovery. Both forms of
training were performed five days a week for six weeks. Moderate-intensity continuous training improved
aerobic power but not maximum O2 debt, while HIT simultaneously improved V
̇O2max, oxygen kinetics and
maximum O2 debt by 28%.
The duration of both work and recovery intervals markedly influences physiological responses.
Olbrecht et al. [3] studied typical interval swimming sets with distances between 50 and 400 m. With 10 s
rest periods, the swimming velocities corresponding to the same blood lactate level as during continuous
swimming increased for 50, 100, 200, and 400 m intervals by 11, 4, 3 and 2% of the velocity at 4 mmol·l-1
(v4), respectively. With 30 s rest, swimming velocity for the 100, 200, and 400 m swims increased in
comparison to v4. On the basis of simulations obtained using mathematical models of energy metabolism
developed by Mader [29], these interactions were confirmed between the aerobic and glycolytic processes
by showing their effects on the individual metabolic responses both during repeated sessions of interval
training and during competition [20, 30]. It appears that the individual metabolic profile (i.e., V
̇O2max for
aerobic potential and V
̇Lamax for glycolytic potential) should be taken into account to individualize training.
For example, it has been shown that World and Olympic-level athletes can performed HIT sets with blood
lactate levels similar to regional level swimmers but at much higher pace. The similar blood lactate levels
for different swimming speeds in the two populations indicates the higher level of O2 uptake associated
with a greater lactate oxidation per unit of time among the fastest swimmers [31]. Swimmers with different
metabolic capacity, aerobic (V
̇O2max) and anaerobic (V
̇Lamax), can exhibit identical speed/lactate
relationships, which is a source of bias for training individualization based only in the speed/lactate curves.
Olbrecht [20] showed that a swimmer with a V
̇O2max of 65 ml·kg-1·min-1 and a V
̇Lamax of 0.40 mmol·l-l·s-1
can have the same lactate/speed curve and the same v4 as a second swimmer with lower V
̇O2max (59 ml·kg-
1·min-1) and a V
̇Lamax (0.20 mmol·l-l·s-1. However, at v4, the first swimmer, with stronger metabolic
potential, will overload less her/his metabolic system compared to the second swimmer. Therefore, the
individual evaluation of a swimmer’s metabolic profile and simulation of their responses to training and
competition should form a rich, reliable and promising method for performance enhancement.
The metabolic responses to different HIT sets with varied speeds, lengths and recovery times based on
the metabolic capacity of swimmers have not studied systematically. However in high-level swimmers it is
difficult to manipulate all HIT parameters in experimental conditions. Furthermore, for the protocols in
which gas exchange is measured continuously, swimmers cannot perform as in real competitive conditions.
Differences in pacing and swimming technique can modify the physiological and metabolic responses.
Finally, the accurate assessment of physiological responses, including the kinetics of important metabolic
parameters such as glycolysis, phosphocreatine (PCr) depletion and muscle pH, is not feasible in real
swimming conditions. Metabolic simulation using mathematical modeling offers an effective solution. The
aim of this study was to compare the simulated metabolic responses to 11 sets of HIT training over 25, 50,
100 and 200 m with a total length of 1000 m and different recovery times in a group of elite swimmers.
METHODS
Participants
Participants were seven male international-level swimmers 21 ± 4 years, 77 ± 8 kg, 188 ± 9 cm. The
intervention consisted of three parts: (i) obtaining a full metabolic performance profile of each swimmer
including aerobic and glycolytic capacity, body composition and energy cost of swimming; (ii) performing
a variety of HIT training sessions utilising different distances, speeds and recovery times, and (iii)
simulating the dynamics of different parameters of energy metabolism for each HIT session, based on the
training load and individual metabolic characteristics of the swimmers.
Each swimmer underwent a series of performance tests in a 50 m pool, comprising a 3 x 400 m
incremental test, an all-out 1 x 100 m test, and an all-out 1 x 50 m test, all swum in front crawl technique.
The intensity of the 3 x 400 m test was: one at low speed, the second at medium speed and the last at
volitional maximal speed. Swimmers were asked to swim at steady speed for each trial and acoustic
feedback on their pace was given every 50 m throughout the test. Both the 1 x 100 m and the 1 x 50 m tests
were performed at volitional maximum speed. Except for the 1 x 50 m test, which was started with a dive
from the blocks, all tests were performed with a push start in the water.
Capillary blood samples (5 µl) were collected from a fingertip and lactate concentration analyzed using
a Lactate Pro 2 handheld device (Arkray, Japan). Three blood samples were taken before each exercise.
Post-exercise samples were taken immediately after each test and then at 1 min intervals until a decline of
lactate values was obtained. Gas exchange was measured (K4 b2, Cosmed, Italy) immediately after each of
the 3 x 400 m swims and the 1 x 100 m swim for 30 s. Body composition was assessed using a
bioimpedance analyzer (Tanita DC 360-S, France).
Data Analysis
The 30 s backward extrapolation method (SEE 3.6%, Chaverri et al. [32]) was used to estimate V
̇O2 for
the 3 x 400 m and the 1 x 100 m tests. For each test, time, distance, lactate and V
̇O2 (where applicable)
were entered into a metabolic analysis software (INSCYD GmbH, Taegerwilen, Switzerland). The software
is an evolution of the algorithms and set of equations developed by Mader [29] with additional procedures
and routines, including the ability to process data derived from field tests. V
̇O2 and lactate data from the 3 x
400 m and the 1 x 100 m tests were used to determine the individual Etot based on the O2 demands as a
function of swimming speed for each subject (figure 1).
Figure 1. Computer simulation of the total energy demands expressed in O2 equivalent (black line) and
oxygen uptake (blue line) as a function of swimming speed in one elite swimmer.
Total energy expenditure (Etot), expressed as total O2 demand (V
̇O2tot), was determined as the sum of
measured V
̇O2 and the oxidative energy equivalent of the accumulated blood lactate during each test (2.5 to
3.0 mlO2·mmol-1·kg-1 depending on the swimmer’s mass according to Di Prampero [33]. At a given
intensity (i.e., swimming speed), the actual lactate production and oxidation rates can be described as a
function of the maximum lactate production rate (V
̇Lamax) and maximum oxidation rate (V
̇O2max),
respectively. The maximum lactate oxidation rate is limited by the actual activation of the aerobic
metabolism and is therefore a function of V
̇O2max [29]. Thus, numerous mathematical solutions for a single
lactate value exist. The combined data of all tests were processed using non-linear optimization functions.
By using conventional mathematical minimization routines of measured vs. calculated lactate
concentrations, the underlying V
̇O2max and V
̇Lamax were estimated using the metabolic analysis software
(Table 1). The swimmers completed a set of HIT swimming workouts of various forms, all covering a total
distance of 1000 m. Table 2 provides an overview of these sessions. The individual average duration of the
interval and recovery periods were recorded for each session.
––Table 2––
To simulate the dynamics of the muscular energy metabolism during each individual interval training
session, the INSCYD metabolic analysis software was used. For each HIT session a load profile was
created. Each profile contained the time-dependent metabolic demands, based on the individual metabolic
requirement vs. speed equation. Metabolic performance variables (V
̇O2max, body composition, V
̇Lamax)
were then used to calculate the dynamics of muscular energy metabolism. Calculation default values were
used for unmeasured parameters. The time constant for the monoexponential phase of V
̇O2 kinetics was 12
s, muscular PCr content at rest was 20 mmol·kg-1, and muscular pH at rest was 7.4.
Statistical Analyses
Mixed-effects models were used to estimate the effects of the individual metabolic profile (e.g.,
V
̇O2max, V
̇Lamax) on the metabolic responses during exercise (e.g., mean V
̇O2, ΔV
̇O2, %V
̇O2max, % energy
contribution). These models are usually used to analyze longitudinal data because they account for both
inter-subject variability and intra-subject correlation. The effects common to the entire cohort were
estimated with the fixed-effects part of the model. We accounted for individual differences in baseline
performance level by including random intercepts. All models were coded for adjustment variables.
Metabolic responses were then compared using the Wilcoxon non-parametric test for paired samples. All
statistical analyses were conducted in R.
RESULTS
The aerobic and anaerobic alactic (Figure 2), anaerobic glycolytic and muscle pH responses (Figure 3)
and metabolic energy contribution from the fours metabolic sources (Figure 4) are displayed for the four
types of HIT series.
Figure 2. Metabolic computer simulation of oxygen uptake (blue), phosphocreatine (green) and power (grey) for
one elite swimmer during HIT series: (upper left) 5 x 200 m (start at 2:30), (upper right) 10 x 100 m (start at
1:20), (bottom left) 20 x 50 m (start 1:00) and (bottom right) 40 x 25 m (start at 0:30). Times are min:s.
Figure 3. Metabolic computer simulation of lactate production rate (pink), phosphocreatine (black) and power
(grey) for one elite swimmer during HIT series: (upper left) 5 x 200 m (start at 2:30), (upper right) 10 x 100 m
(start at 1:20), (bottom left) 20 x 50 m (start 1:00) and (bottom right) 40 x 25 m (start at 0:30). Times are min:s.
Figure 4. Simulated relative metabolic energy contribution (%Etot) from the aerobic (blue), anaerobic alactic
(red) and anaerobic lactic (green) metabolic energy sources for one elite swimmer during HIT series: (upper left)
5 x 200 m (start at 2:30), (upper right) 10 x 100 m (start at 1:20), (bottom left) 20 x 50 m (start 1:00) and (bottom
right) 40 x 25 m (start at 0:30). Times are min:s.
The effects of the independent variables V
̇O2max and V
̇Lamax and exercise covariates (swimming speed,
interval length and recovery time) on the dependent aerobic variables (mean V
̇O2, ΔV
̇O2, %V
̇O2max and %
aerobic energy contribution and on anaerobic alactic parameters (mean PCr, ΔPCr and % anaerobic alactic
energy contribution) are shown in Table 3 and 4 (respectively). The coefficients represent the effect on the
dependent metabolic variable (increase or decrease if the coefficient is positive or negative, respectively, with
the other variables assumed to be constant. For instance (Table 3), the effects of the V
̇Lamax on the average V
̇O2
during the different sets of interval training was estimated in the following manner: mean V
̇O2 = -14.8-26 ·
V
̇Lamax. According to this equation, the higher V
̇Lamax, the lower V
̇O2 during the various interval training series.
In Table 4, the effects of speed on PCr changes during the different sets of interval training was estimated as
follows: mean ΔPCr = 8.9 · speed. According to this equation, the higher the speed, the lower the PCr during the
various interval training series.
Table 5 shows the effects of V
̇O2max and exercise covariates on the dependent anaerobic lactic
variables (mean V
̇La, %V
̇La/V
̇Lamax, and % anaerobic lactic energy contribution). The effects of the V
̇Lamax
on the average V
̇O2 during the different sets of interval training was estimated, for example, as follows: mean
V
̇O2 = -14.8-26 · V
̇Lamax. According to this equation, the higher V
̇Lamax, the lower V
̇O2 during the various
interval training series.
––Table 3––
––Table 4––
––Table 5––
Tables 6 to 9 show the differences in selected metabolic parameters and %Etot contribution for each
type of training intervals: 5 x 200, 10 x 100, 20 x 50 and 40 x 25 m (respectively) with various recovery
intervals (short, medium and long). For the 5 x 200 m intervals (Table 6), longer recovery periods were
associated with faster swimming speeds (m·s-1): 1.55 (1.46-1.64; mean, 95% confidence interval, p <0.05),
1.57 (1.50-1.64) and 1.52 (1.42-1.61) for long, medium and short recovery, respectively, which
corresponded to 97, 99 and 100% of v400 (respectively). M ean V
̇O2, ΔV
̇O2, mean %V
̇O2max, mean V
̇La,
ΔV
̇La, %V
̇Lamax, ΔpH, PCr and % anaerobic energy contribution, both alactic and lactic, were greater
when recovery was longer and speed was higher. In contrast, mean PCr and % aerobic contribution
decreased.
––Table 6 to 9––
Likewise, longer recovery periods for the 10 x 100 m intervals (Table 7) elicited faster swimming
speeds (m·s-1): 1.61 (1.51-1.70), 1.65 (1.55-1.75) and 1.68 (1.60-1.76) for long, medium and short
recovery, corresponding to 102, 105 and 107% v400 (respectively). ΔV
̇O2, mean V
̇La, ΔV
̇La, %V
̇Lamax,
ΔpH, mean PCr and % anaerobic alactic and lactic energy contribution were higher when recovery was
longer and speed faster, whereas % aerobic energy contribution decreased. As an example of inter-subject
variability, Figure 5 shows the computer simulation of power, V
̇O2, and PCr for four elite swimmers during
10 x 100 m intervals (start at 1:20). Figure 6 shows the metabolic energy share for the same interval series
with different recovery times.
Figure 5. Metabolic computer simulation of the time course of oxygen uptake (blue), phosphocreatine (green)
and power (grey) for four elite swimmers during 10 x 100 m intervals (start at 1:20). Times are min:s.
Figure 6. Simulated metabolic energy contribution (%Etot) from the aerobic (blue), anaerobic alactic (red) and
anaerobic lactic (green) metabolic energy sources for one elite swimmer during 10 x 100 m intervals (start at
1:20, 1:30 and 1:45, respectively from left to right). Times are min:s.
Similarly for the 20 x 50 m intervals (Table 8), longer recovery periods were associated with faster
swimming speeds (m·s-1): 1.72 (1.64-1.81, mean, 95% confidence interval), 1.80 (1.71-1.89), 1.84 (1.76-
1.91) for long, medium and short recovery (110, 112 and 115% v400, respectively). ΔV
̇O2, mean %V
̇O2max,
mean V
̇La, ΔV
̇La, %V
̇Lamax and % anaerobic alactic contribution were greater when recovery was longer
and speed was faster, whereas % aerobic energy contribution decreased. Finally, for the 40 x 25 m intervals
(Table 9), longer recovery periods elicited faster swimming speeds (m·s-1): 2.00 (1.92-2.09) and 2.07 (1.99-
2.15) for long and short recovery (128 and 132% v400, respectively). Mean V
̇O2, ΔV
̇O2, mean %V
̇O2max,
mean V
̇La, ΔV
̇La, %V
̇Lamax, ΔpH, ΔPCr and % anaerobic alactic and lactic contribution were greater when
recovery was longer and speed was faster, while % aerobic contribution decreased.
Concerning the comparison of HIT series with short recovery, mean V
̇La, ΔV
̇La, %V
̇Lamax, ΔpH,
ΔPCr and both % alactic and lactic contribution were lower in 5 x 200 m with medium recovery (~15-25 s,
1.55 m·s-1, 99% v400) compared with 10 x 100 m with short recovery (~15-20 s, 1.61 m·s-1, 102% v400).
Conversely, mean pH, mean PCr and % aerobic contribution were greater. Compared with 20 x 50 m
intervals with short recovery (~15 s), during 5 x 200 m, ΔPCr and % lactic contribution were smaller
whereas mean pH and % aerobic were higher. Finally, during 10 x 100 m with short recovery (~15-20 s), %
aerobic contribution was higher while ΔV
̇O2 and % anaerobic contribution were lower.
Regarding the comparison of HIT series with medium recovery, ΔpH, ΔPCr, %V
̇Lamax, and % alactic
and lactic contribution were lower for 5 x 200 m intervals with medium recovery (~25-35 s) compared with
10 x 100 with similar recovery (~25-30 s). In contrast, % aerobic contribution was greater. In 5 x 200 m
intervals compared to 20 x 50 m series with medium recovery (~30 s), mean V
̇La, %V
̇Lamax, ΔpH, mean
PCr, ΔPCr and % anaerobic lactic contribution were lower, but % aerobic was greater. Similarly, in 5 x200
m intervals with medium recovery (~25-35 s), compared with 40 x 25 m with similar rest, ΔV
̇O2, mean
V
̇La, ΔV
̇La, %V
̇Lamax and % lactic contribution were lower, while % aerobic contribution was greater.
With regard to 10 x 100 s compared with 20 x 50 m intervals, ΔV
̇O2, ΔV
̇La and % anaerobic lactic
contribution were higher and % aerobic contribution was greater. Finally, 20 x 50 m compared with 40 x 25
m intervals with medium recovery evoked lower ΔV
̇O2, mean %V
̇O2max, V
̇La and lactic contribution, but
greater % alactic energy share.
Finally, when referring to the comparison of HIT series with long recovery, we have observed that,
compared with 10 x 100 m intervals with long recovery (~35-40 s), 5 x 200 m series elicited lower ΔV
̇O2,
mean V
̇La, %V
̇Lamax, ΔpH, ΔPCr and % alactic and lactic contribution, while % aerobic contribution was
greater. For 5 x 200 m compared with 20 x 50 m intervals with long recovery (~45 s), mean V
̇La,
%V
̇Lamax, mean PCr, ΔPCr and % alactic and lactic contribution were lower, but mean V
̇O2, mean
%V
̇O2max and % aerobic contribution were higher. Similarly, during 5 x 200 m compared with 40 x 25 m
intervals with long recovery (~30 s), ΔV
̇O2, mean PCr, ΔPCr and % lactic share were lower, while mean
V
̇O2, mean %V
̇O2max and % aerobic contribution were greater. During 10 x 100 m, compared with long-
recovery 20 x 50 m series, mean V
̇O2, mean %V
̇O2max and % aerobic contribution were higher, while mean
PCr and lactic contribution were lower. Similarly, 10 x 100 m compared with 40 x 25 m intervals evoked
lower ΔV
̇O2, mean pH and % alactic and lactic contributions, but higher mean V
̇O2, mean %V
̇O2max and %
aerobic contribution. Finally, 20 x 50 m compared with 40 x 25 m intervals, caused lower ΔV
̇O and %
lactic contribution but higher % aerobic and alactic energy share. Finally, during the long-recovery 10 x
100 m intervals and the short-recovery 20 x 50 m series the swimmers reached a similar speed (1.68 m·s-1,
107% v400 vs. 1.72 m·s-1, 110% v400, respectively). ΔV
̇O2, ΔV
̇La and % lactic contribution were lower
for the 10 x 100 m series, whereas mean V
̇La, %V
̇Lamax and % alactic contribution were higher.
DISCUSSION
The simulation of metabolic responses to various HIT sessions in six male elite swimmers, ranging
between 94 and 132% v400 (~vV
̇O2max) forms a new set of useful results for training programming. For
aerobic metabolic parameters, multivariate models showed that mean V
̇O2 was higher when V
̇O2max and
swimming speed were high, work intervals long and recovery intervals short. Mean V
̇O2 during HIT
sessions was lower when V
̇O2max was low, V
̇Lamax and speed were high, and recovery long. For example,
mean V
̇O2, V
̇O2max and % aerobic contribution was highest for the 100 and 200 m intervals (both using
~20-45 s recovery) than 50 m (~14-45 s recovery) and 25 m intervals (~15-30 s recovery).
For indices of anaerobic lactic metabolism, mean V
̇La in absolute values and in %V
̇Lamax were higher
for greater V
̇Lamax and faster swimming speed. In turn, anaerobic lactic contribution was greater when
V
̇O2max was low, swimming speed fast, work intervals short and recovery time long. For instance, mean
V
̇La, %V
̇Lamax were greater in the 50 and 100 m intervals with longer recovery (~45 s). Regarding the
anaerobic alactic parameters, PCr concentration was highest for short work intervals (25 and 50 m), and its
depletion over time was greater when V
̇Lamax and speed were higher. A large contribution of the anaerobic
alactic metabolism was associated with lower V
̇Lamax. For all intervals studied (25, 50, 100 and 200 m),
longer recovery was associated with faster swimming speed, inducing greater mean V
̇O2 and glycolytic
supply. The aerobic contribution was lower while the anaerobic lactic share was greater.
Mean V
̇O2 during the different sets of interval training (in absolute values) was even higher than V
̇O2
measured during the maximal 400 m test, which has been considered as swimming-specific V
̇O2max [34].
The Mean V
̇O2 increased with swimming speed, longer work intervals and shorter recovery periods. Our
results showed an increase in speed across the 200, 100 and 50 m intervals (97-100, 102-105 and 110-
117% vV
̇O2max, respectively). Swimming scientists should work with coaches (and swimmers) to convert
these standardized effects (percentages) to absolute times (min:sec) that is more familiar to the high
performance swimming community.
Mean V
̇O2, %V
̇O2max and the aerobic contribution were higher during the 200 m intervals (80-85%, 60-
75% and 35-65% Etot, respectively). In running it appears that prolonging exercise duration increases the
relative aerobic energy requirements [35] as well as the time spent close to V
̇O2max [36]. In the case of high
intensities equivalent in power or speed, lengthening work intervals increased both the contribution of
glycogenolysis and the total energy demand [5]. However, this shift to a larger anaerobic share was not
observed in our study, likely because of a gradual decrease in swimming speed in the longer intervals
(132% and 97% v400 for 25 m and 200 m intervals, respectively).
In elite swimmers, Sousa et al. [24] estimated higher aerobic contribution (83%) during a time limit
test (square-wave exercise lasting ~344 s) swum at 95% vV
̇O2max compared to 100% vV
̇O2max (~194 s,
74% aerobic) and 105% vV
̇O2max (~123 s, 59% aerobic). The greater aerobic contribution during 200 m
intervals can be explained by the total duration of the effort, since the aerobic contribution is calculated as
the integral over time of the V
̇O2-time relationship. If we compare our results with those of Souza and co-
workers we observe a greater aerobic contribution while the duration of the 200 m swimming intervals in
our study were close to the time limit swims in the cited study (80-85% aerobic for ~130 s of exercise in
this study versus 59% aerobic in Souza’s study for time limit tests of ~123 s). Although swimming speed
was higher in our study for the 200 m intervals, the total effort was distributed over a total distance of 1000
m: 1.5-1.6 m·s-1 (92-96% v400) compared with 1.3 m·s-1 for the cited study, representing a time limit
achieved to the maximum capacity of the swimmers (105% vV
̇O2max).
Care should be taken in comparing the results of this study, obtained by a simulation method, and
others based on experimental measurements. Mean V
̇O2 became lower when the swimmer’s V
̇O2max was
lower, V
̇Lamax and speeds were high, and recovery times were longer. Previous studies that have analysed
competitive events at maximum intensity reported a decline in V
̇O2 at the end of maximal running trials
and in swimming [12, 37, 38]. This decline has been interpreted as an inhibition of mitochondrial activity
caused by excessive acidosis and feedback glycolysis inhibition or by depletion of PCr stores. Our
simulation results support this contention [28] by showing that high V
̇Lamax associated with low V
̇O2max
leads to a gradual decline of mean V
̇O2 and PCr concentrations, even in submaximal speeds such as those
used in this study. Therefore, to maximize aerobic demands coaches should use long intervals of 200 and a
100 m swum between 95% and 110% v400 with short, medium and long recovery periods (10 to 45 s) for
200 m intervals, and short and medium recovery for 100 m intervals (15 to 25 s). With the aim of HIT
individualization and to maintain the highest possible V
̇O2 series, swimmers with high V
̇O2max and low
V
̇Lamax should preferentially use long recovery after the fastest possible speed, whereas swimmers with low
V
̇O2max and high V
̇Lamax should reduce swimming speed and use short recovery.
For indices of anaerobic glycolytic metabolism, mean V
̇La – both in absolute values and as % V
̇Lamax –
were higher in those swimmers who exhibited greater V
̇Lamax and swam faster. The simulation results
shows that mean V
̇La were highest during medium-length intervals (50 and 100 m), with the longest
recovery and, thus, at fastest speeds. In interpreting these results, three explanations are possible. First, the
anaerobic glycolytic system is fully solicited during the 50 and 100 m intervals because its maximum
power is reached after 10 to 30 s of exercise [10]. Secondly, swimming speeds were higher in the 50- and
100 m series compared to the 200 m intervals (102-117 vs. 97-100% vV
̇O2max, respectively). Finally, the
rapid decrease in the oxidative rate occurring during passive rest reduces the aerobic contribution at the
beginning of the subsequent intervals. Moreover, the contribution of the fast glycolytic system to energy
supply increases rapidly after the initial 10 s of intense exercise. By about 20 s of sustained intense activity
most of the total energy is supplied by the anaerobic glycolytic system. By 45 s, the decline in power
output relates to a reduced energy production of the glycolytic system. Beyond this point a growing
reliance on the aerobic system occurs [10]. Exercise duration associated with 50 and 100 m intervals (~30
and ~60 s, respectively) correspond to the highest energy contribution of the glycolytic system. In our
study, for a given distance, higher glycolytic output was observed for the longest recovery intervals (i.e.,
~40 vs. ~15 s for the 100 m and ~15 vs. ~30 s for the 50 m). Longer recovery periods were associated with
higher speeds confirming previous studies in running in which 15 to 60 s work intervals elicited higher
blood lactate levels [2]. With longer recovery intervals, the larger glycolytic reliance could also be related
to the lower V
̇O2 reached during longer recovery periods, which could increase the O2 deficit at the
beginning of the subsequent work interval [39].
During 50 and 100 m series with long rest, pH reached the lowest levels, which could explain the
greater V
̇O2 decline during the 20 x 50 m / 45 s recovery series. This hypothesis is consistent with studies
in other sports showing that during work at intensities ≥ 120% vV
̇O2max, the exercise capacity could be
impaired and, hence, the total time close to V
̇O2max can be substantially reduced [40]. This type of 50 and
100 m intervals at full speed could be used to maintain a good technique despite the increase in muscle
fatigue. This type of workload may be suitable for high-level swimmers who are able to achieve HIT
sessions at this intensity for ≥ 8 min [41]. For the sake of training individualization, 50 and 100 m intervals
swum at the highest possible speed with long recovery may be recommended to 50 and 100 m swimmers
with very high V
̇O2max and low V
̇Lamax.
In our study, 25 m intervals at speed equivalent to 128% v400max with short recovery evoked medium
V
̇O2 levels (and %V
̇O2max) close to the highest values reached during 100 and 200 m intervals (61 ml·kg-
1·min-1 and 81% V
̇O2max vs. 63 ml·kg-1·min-1 and 85% V
̇O2max, respectively) despite being swum at much
lower speed (100 and 107% v400, respectively). Moreover, during 25 m intervals at very fast speed, the
glycolytic activation and muscle pH values were similar to those observed in the 50 and 100 m intervals at
slower speed, confirming the effectiveness of this type of series as a stimulus for the cardiovascular system
associated with very fast speed and good technical quality [2]. In rowing, this 15/15 model showed a
relatively high aerobic loading with low glycolytic activity [42]. It appears this type of HIT is as an
alternative model for training to allow rowers to realize substantial amounts of training at intensities
slightly higher than those of competition. The results of our simulations also indicate that mobilization of
the anaerobic alactic system (PCr concentration, PCr decrease over time) was important during short
intervals and for swimmers exhibiting low V
̇Lamax. Therefore, 25 m series with short recovery should
facilitate joint mobilization of the aerobic and anaerobic alactic metabolism close to those observed in 100
and 200 m races with matching technical requirements. By lengthening recovery from 15-20 to 30-35 s, 40
x 25 m HIT intervals permitted 4-6% faster speeds, 25-30% increased glycolytic power and 45-60% higher
anaerobic lactic contribution. This type of long-recovery 25 m intervals could allow workout speeds and
stroke frequencies associated with mixed aerobic and anaerobic lactic energy supply in close proportion to
those observed in the 100 m events.
Regarding 50, 100 and 200 m intervals, the longer the recovery and the higher the speed and the
glycolytic activation, the more stable was V
̇O2 and the lower the aerobic energy share, but the anaerobic
glycolytic and alactic contributions were greater. These results reflect the design of our protocol in which
longer recovery time was systematically associated with the elevation of swimming speed. In the case of
running, previous work already highlighted that for short HIT series, higher work interval intensities
elicited higher blood lactate levels [4]. In interval training at supramaximal intensity different metabolic
sources may be activated depending on the recovery duration. For example, interval training leads to lower
blood lactate concentrations for similar or even higher V
̇O2 levels than those measured during longer
interval training or continuous training performed at similar intensity [3, 43, 44]. For identical constant
speed, exercise time before exhaustion can increase three-fold by including 10 s recovery periods and six-
fold by using for 20 s recovery periods. Furthermore, blood lactate accumulation stabilized at 11, 7.5 and 2
mmol·l-1 for 10, 20 and 30 s recovery intervals, respectively [44]. Lengthening the duration of recovery in
the case of training bouts realized at similar intensity enables the restitution of myoglobin-bound oxygen
stores and levels of ATP and PCr [45]. During short interval training, Medbø et al. [46] reported that a
large portion of O2 was stored in muscle myoglobin during resting periods (10% of the maximal
accumulated O2 debt, MAOD), minimizing depletion during the periods of exercise and thus relying less on
the glycolytic pathway during exercise. The fall in PCr during exercise would be followed by resynthesizes
during recovery [47] leading to lesser accumulation of lactate in the muscles and lower V
̇O2 drop compared
with continuous training or longer interval sets. In summary, when swimmers are asked to achieve their
best average times, lengthening the recovery time allows them to swim faster thanks to the attainment of
similar V
̇O2max levels (or slightly higher in the case of 200 m intervals) and use of their anaerobic
glycolytic power.
CONCLUSION
This pioneering study was designed to analyze the metabolic responses to HIT using computer
simulation based on the mathematical modeling of energy metabolism (Mader 2003) during swimming. For
that purpose, six elite swimmers were first tested in the pool for metabolic capacities and then we simulated
their metabolic response during different HIT series over 200, 100, 50 and 25 m with short, medium and
long recovery time, swum to the best possible average times. The aerobic responses during HIT were
greatest for swimmers with the strongest aerobic potential and low glycolytic capacity over 200 and 100 m
intervals with short recovery. The highest glycolytic energy supply and the lowest muscle pH were
observed in swimmers with the strongest glycolytic power and during 100 and 50 m intervals with longer
recovery. Shorter 25 m intervals were swum at the fastest speed and with high aerobic energy output,
balanced glycolytic output during shorter recovery and predominantly glycolytic output for longer recovery
periods. For all types of simulated intervals, the lengthening of the recovery periods increased oxidative
and glycolytic expenditure, moving energy contribution to a predominance of anaerobic sources.
REFERENCES
1. Buchheit, M., Laursen, PB (2013a). High intensity interval training, solutions to the programming puzzle.
Part I: Cardiopulmonary emphasis. Sports Med, 43(5), 313-38.
2. Buchheit, M., Laursen, PB (2013b). High intensity interval training, solutions to the programming puzzle.
Part II: Neuromuscular load and practical application. Sports Med, 43(10):927-54.
3. Olbrecht, J., Madsen, O., Mader, A., Liesen, H., Hollmann, W. (1985). Relationship between swimming
velocity and lactic concentrations during continuous and intermittent training exercises. Int J Sports Med, 6,
74-77.
4. Billat, L.V. Interval training for performance: A scientific and empirical practice. Special recommendations
for middle-and long-distance running (2001). Part I: Aerobic Interval Training. Sports Med, 31(1), 13-31.
5. Billat, L.V. Interval training for performance: A scientific and empirical practice. Special recommendations
for middle-and long-distance running (2001). Part II: Anaerobic Interval Training. Sports Med, 31(2), 75-
90.
6. Seiler, S., Sjursen, J.E. (2004). Effect of work duration on physiological and rating scale of perceived
exertion responses during self-paced interval training. Scan J Med Sci Sports, 14:318-25.
7. di Prampero, P.E., Pendergast, D.R., Wilson, D.W., and Rennie, D.W. (1978). Blood lactic acid
concentrations in high velocity swimming, in Swimming Medicine IV, eds. B. Eriksson & B. Furberg.
(Baltimore: University Park Press), 249-261.
8. Capelli, C., Zamparo, P., Cigalotto, A., Francescato, MP., Soule, R.G., Termin, B., Pendergast, D.R., Di
Prampero, P.E. (1995). Bioenergetics and biomechanics of front crawl swimming. J Appl Physiol,
78(2):674-9.
9. Toussaint, H.M., Hollander, A.P. (1994). Energetics of competitive swimming. Sports Medicine.
1994;18(6):384-405.
10. Rodríguez, F.A., Mader, A. Energy systems in swimming. In: Seifert L, Chollet D, Mujika I, eds. World
Book of Swimming: From Science to Performance. Hauppauge, New York: Nova Science Publishers;
2011:225-240.
11. Capelli, C., Pendergast, D.R., Termin, B (1998). Energetics of swimming at maximal speeds in humans.
Eur J Appl Physiol Occup Physiol, 78: 385-393.
12. Hellard, P., Pla, R., Rodriguez ,F.A., Simbana, D., Pyne, D.B (2018). Dynamics of the metabolic response
during a competitive 100-m freestyle in elite male swimmers. Int J Sports Physiol Perform, 21:1-28
13. Hellard, P., Avalos, M., Lacoste, L., Barale, F., Chatard, JC., Millet, GP (2006). Assessing the limitations
of the Banister model in monitoring training. J Sports Sci, 24(5):509-20.
14. Vandenbogaerde, T., Hellard, P., Derave, W. (2018). Swimming high intensity interval training. In
Buccheit, M., Laursen, P. Science and application of high intensity interval training. Human Kinetics
(Eds.).
15. Oliveira MF, Caputo F, Dekerle J, Danadai BD, Greco CC. Stroking parameters during continuous and
intermittent exercise in regional-level competitive swimmers (2012). Int J Sports Med, 33(9):696-701.
16. Dekerle, J., Nesi, X., Lefevre, T., Depretz, S., Sidney, M., Marchand, F.H., Pelayo, P (2005). Stroking
parameters in front crawl swimming and maximal lactate steady state speed. Int J Sports Med, 26(1):53-8.
17. Dekerle, J., Paterson, J. Muscle fatigue when swimming intermittently above and below critical speed.
(2016). Int J Sports Physiol Perform, 11(5):602-16.
18. Hellard, P., Dekerle, J., Nesi, X., Toussaint, J.F., Houel., N., Hausswirth, C. (2010). Modelling the VO2
slow component in elite long distance swimmers at the velocity associated with lactate threshold. In: P.L.
Kjendlie, R.K. Stallman, Cabri, J (Eds). Biomechanics and Medicine in Swimming XI, (pp.196-198).
Norwegian School of sport Sciences.
19. Bentley, D.J., Roels, B., Hellard, P., Fauquet, C., Libicz, S., Millet, G.P. (2005). Physiological responses
during submaximal interval swimming training: effects of interval duration. J Sci Med Sport, 8(4), 392-402.
20. Olbrecht, J (2007). The Science of Winning. Planning, Periodizing and Optimizing Swim Training.
Antswerp, F&G Partners.
21. Costill, D.L., Flynn, M.G., Kirwan, J.P., Houmard, J.A., Mitchell, J.B., Thomas, R., Park, S.H. (1988).
Effects of repeated days of intensified training on muscle glycogen and swimming performance. Med Sci
Sports Exerc, 20(3):249-54.
22. Mujika, I., Chatard, J.C., Busso, T., Geyssant, A (1995). Effects of training on performance in competitive
swimming. Can J Appl Sport Sci, 20:395-406.
23. Termin, B., & Pendergast, D. R. (2000). Training using the stroke frequency-velocity relationship to
combine biomechanical and metabolic paradigms. Journal of Swimming Research, 14, 9-17.
24. Sousa, A.C., Vilas-Boas, J.P., Fernandes, R.J. V
̇O2 kinetics and metabolic contributions whilst swimming at
95, 100 and 105% of the velocity at V
̇O2 max. Biomed Research International, 675363.
25. Sousa, A.C., Vilas-Boas, J.P., Fernandes, R.J., Figueiredo, P. (2016). VO2 at Maximal and Supramaximal
Intensities: Lessons to High Interval Training in Swimming. Int J Sports Physiol Perform, 12(7):872-877.
26. Baker, J.S., McCormick, M.C., Robergs, R.A. (2010). Interaction among Skeletal Muscle Metabolic
Energy Systems during Intense Exercise. J Nutr Metab. 905612.
27. Ogita, F. Energetic in competitive swimming and its application for training. In: Vilas-Boas JP, Alves F,
Marques A. Biomechanics and Medicine in Swimming X. Revista Portuguesa de Ciencias do Desporto;
2006;6(2) :117-121.
28. Ogita, F. (2011). Training Energy Systems. World Book of Swimming: From Science to Performance.
Hauppauge, New York: Nova Science Publishers; 2011:241-255.
29. Mader, A. Glycolysis and oxidative phosphorylation as a function of cytosolic phosphorylation state and
power output of the muscle cell (2003). Eur J Appl Physiol. 88(4-5):317-38.
30. Weber, S., Hellard, P., Rodríguez, F.A., Mader, A. Metabolic profiling in elite swimmers – testing in the
pool to determine aerobic & glycolytic capacities. 18th FINA World Sports Medicine Congress, 2016.
31. Mader, A., Heck, H. A theory of the metabolic origin of "anaerobic threshold" (1986). Int J Sports Med. 7
Suppl 1:45-65.
32. Chaverri, D., Iglesias, X., Schuller, T., Hoffmann, U., and Rodríguez, F.A. (2016). Estimating peak oxygen
uptake based on postexercise measurements in swimming. Appl Physiol Nutr Metab 41(6), 588-596.
33. Di Prampero, P.E., Pendergast, D.R., Wilson, D.W., Rennie, D.W. Energetics of swimming in man. J Appl
Physiol. 37(1):1-5.
34. Rodríguez, F.A. (2000). Maximal oxygen uptake and cardiorespiratory response to maximal 400-m free
swimming, running and cycling tests in competitive swimmers. J Sports Med Phys Fitness 40(2), 87-95.
35. Gastin, P.B. (2001). Energy system interaction and relative contribution during maximal exercise. Sports
Med. 31:725-741.
36. Rozenek, R., Funato, K., Kubo, J., Hoshikawa, M., Matsuo, A. (2007). Physiological responses to interval
training sessions at velocities associated with VO2max. J Strength Cond Res. 21(1):188-92.
37. Hanon, C., Lepretre, P.M., Bishop, D., Thomas, C. (2010). Oxygen uptake and blood metabolic responses
to a 400-m run. Oxygen uptake and blood metabolic responses to a 400-m run. Eur J Appl Physiol.
109(2):233-40.
38. Jalab, C., Enea, C., Delpech, N., Bernard, O. (2011). Dynamics of oxygen uptake during a 100 m front
crawl event, performed during competition . Appl Physiol Nutr Metab. 36(2):219-225.
39. Rusko, H.K. (1992). Development of aerobic power in relation to age and training in cross-country skiers.
Med Sci Sports Exerc. 24(9):1040-7.
40. Dupont, G., Blondel, N., Lensel, G., Berthoin, S. Critical velocity and time spent at a high level of VO2 for
short intermittent runs at supramaximal velocities. Can J Appl Physiol. 27(2):103-15.
41. Buchheit, M., Lepretre, P.M., Behaegel, A.L., Millet, G.P., Cuvelier, G., Ahmaidi, S. (2009).
Cardiorespiratory responses during running and sport-specific exercises in handball players. J Sci Med
Sport. 12(3):399-405.
42. Gullstrand, L. (1996). Physiological responses to short-duration high-intensity intermittent rowing. Can J
Appl Physiol 21(3), 197-208.
43. Turner, A.P., Cathcart, A.J., Parker, M.E., Butterworth, C., Wilson, J., Ward, S.A. (2006). Oxygen uptake
and muscle desaturation kinetics during intermittent cycling. Med Sci Sports Exerc. 38(3):492-503
44. Margaria, R., Oliva, R.D., Di Prampero, P.E., Cerretelli, P. (1969). Energy utilization in intermittent
exercise of supramaximal intensity. J Appl Physiol. 26(6):752-6.
45. MacDougall, D., Sale, D. (1981) Continuous vs. interval training: a review for the athlete and the coach.
Can J Appl Sport Sci 6(2):93-97
46. Medbo, J.I., Mohn, A.C., Tabata, I., Bahr, R., Vaage, O., Sejersted, O.M. (1992). Anaerobic capacity
determined by maximal accumulated O2 deficit. J Appl Physiol 73(3):1207-1209
47. Essen, B., Hagenfeldt, L., Kaijser, L. (1977). Utilization of blood-borne and intramuscular substrates
during continuous and intermittent exercise in man. J Physiol, 265(2):489-506
Table 1. Swimming speeds and lactate values for each swimmer.
Table 1. Swimming speeds and lactate values for each swimmer.
S
50-m
(m·s-1)
[La]b,net
(mmol·l-1)
100-m
(m·s-1)
[La]b,net
(mmol·l-1)
400-m
(1st)
[La]b,net
(mmol·l-1)
400-m
(2nd)
[La]b,net
(mmol·l-1)
400-m
(3rd)
[La]b,net
(mmol·l-1)
1
2.12
7.2
1.96
18.8
1.39
1.6
1.49
4.1
1.54
15.9
2
1.96
7.2
1.85
18.8
1.39
1.6
1.49
4.1
1.54
11
3
2.11
11.8
1.89
8.9
1.31
1.6
1.42
3
1.44
10.5
4
2.01
14.1
1.86
15.4
1.32
1.2
1.40
1.6
1.52
10.6
5
1.94
8.7
1.72
9.0
1.41
1
1.46
1.9
1.57
14.8
6
2.13
9.0
1.96
14.3
1.45
2
1.50
2.8
1.57
9.2
7
1.97
10.1
1.83
10.1
1.42
1.5
1.53
3.3
1.58
10.2
Table 2. Swimming speed in absolute and relative values (% v400max) for the different HIT training workouts.
400 m
max
Lactate
thresho
ld[2]
5x200
2:50
5x200
2:40
5x200
2:30
10x100
1:45
10x100
1:30
10x100
1:20
20x50
1:15
20x50
1:00
20x50
0:45
40x25
0:45
40x25m
0:30
Speed
(m·s-1)
1.57
1.50-
1.64
1.48
1.46-
1.49
1.57
1.50-
1.64
1.55
1.46-
1.64
1.52
1.42-
1.61
1.68
1.60-
1.76
1.65
1.55-
1.75
1.61
1.51-
1.70
1.84
1.76-
1.91
1.8
1.71-
1.89
1.72
1.64-
1.81
2.07
1.99-
2.15
2.00
1.92-
2.09
Relative
speed
(%v400max)
100
94
100
99
97
107
105
102
117
115
110
132
128
Data are mean and range. Recovery times are expressed as min:s. All differences between speeds are significant (p <0.05)
Table 3. Effects of maximal oxygen uptake (V
̇O2max), maximal lactate production rate (V
̇Lamax) and high-
intensity interval training (HIT) covariates (swimming speed, interval length and recovery time) on aerobic
parameters (V
̇O2, ΔV
̇O2, %V
̇O2max and % aerobic energy contribution). Interval lengths are 200, 100, 50 and 25
m. Recovery time is a qualitative variable: short, medium, long. See text for further explanations.
Mean V
̇O2
ΔV
̇O2
% V
̇O2/ V
̇O2max
Aerobic energy (% Etot)
Coefficient
(IC 95%)
t
p
F
p
Coefficient
(IC 95%)
t
p
F
p
Coefficient
(IC 95%)
t
p
F
p
Coefficient
(IC 95%)
t
p
F
p
Intersection
-14.8
(-41:12)
0.27
0.53
<0.01
32.4
(14.3-50.6)
<0.01
0.65
<0.01
0.58
(0.22-0.93)
<0.01
0.34
<0.01
126
(98:153)
<0.01
0.94
<0.01
V
̇O2max (ml·kg-1·min-1)
0.93
(0.66:1.17)
<0.01
-0.18
(-0.32:-0.01)
<0.01
0.002
(-0.001:-0.005)
0.26
0.38
(0.12:0.64)
<0.01
V
̇Lamax (mmol·l-l·s-1)
-26
(-40:-12)
<0.01
10.5
(0.0:0.32)
<0.01
-0.36
(-0.54:-0.17)
<0.01
2.48
(-0.9:-17)
0.73
Speed (m·s-1)
9.7
(-0.2:19)
<0.05
-19.2
(-25:-12)
<0.01
0.13
(0.003:0.26)
<0.01
-55
(-65:-44)
<0.01
Intervals (m)
0.04
(0.01:0.07)
<0.01
0.004
(-0.01:0.02)
0.6
0.0006
(0.0003-0.0009)
<0.01
0.10
(0.07:0.12)
<0.05
Recovery (s)
-1,2
(-2.6:0.25)
<0.01
2.23
(1.24:3.22)
<0.01
-0.02
(-0.003:-0.001)
<0.01
-2.5
(-4:-1)
<0.01
The p-values associated with a single coefficient correspond to Student's t-tests whereas the p-values associated with
the overall model correspond to ANOVA F-tests.
Table 4. Effects of maximal oxygen uptake (V
̇O2max), maximal lactate production rate (V
̇Lamax) and exercise
covariates (swimming speed, interval length and recovery time) on anaerobic alactic parameters (mean PCr,
ΔPCr and % anaerobic alactic contribution). See text for further explanations.
Mean PCr
ΔPCr
Anaerobic alactic energy (%Etot)
Coefficients
(IC 95%)
t
p
F
p
Coefficients
(IC 95%)
t
p
F
p
Coefficient
(IC 95%)
p
Intersection
7.7
(-6.2:21.6)
0.27
0.27
<0.01
-9.5
(-22:3)
0.77
0.53
<0.01
8.6
(-4.5:21.6)
0.2
V
̇O2max (ml·kg-1·min-1)
-0.007
(-0.13:0.12)
0.91
0.04
(-0.07:0.16)
0.64
-0.07
(-0.2:0.04)
0.2
V
̇Lamax (mmol·l-l·s-1)
4.3
(-2.9:11.6)
0.24
-9.7
(-16.3:-3.1)
<0.01
-6.1
(-13:0.9)
<0.05
Speed (m·s-1)
0.04
(-5.1:5.2)
0.98
8.9
(4.2:13.5)
<0.01
0.37
(-2.6:7.1)
0.37
Intervals (m)
-0.01
(-0.03:-0.02)
<0.05
-0.005
(-0.01:0.007)
0.86
-0.32
(-0.006:0.02)
0.32
Recovery (s)
0.6
(-0.2:1.33)
0.13
-0.5
(-1.2:0.17)
0.23
2.2
(-0.4:1.1)
0.35
The p-values associated with a single coefficient correspond to Student's t-tests whereas the p-values associated with
the overall model correspond to ANOVA F-tests.
Table 5. Effects of maximal oxygen uptake (V
̇O2max) and exercise covariates (swimming speed, interval length
and recovery time) on anaerobic lactic metabolic parameters (mean V
̇La, %V
̇La/V
̇Lamax, and % anaerobic lactic
energy contribution). See text for further explanations.
Mean V
̇La
%V
̇La/V
̇Lamax
Anaerobic lactic energy (%Etot)
Coefficients
(IC 95%)
t
p
F
p
Coefficients
(IC 95%)
t
p
F
p
Coefficient
(IC 95%)
t
p
Intersection
-1.35
(-5.7:3.1)
0.54
0.40
<0.01
1.4
(-8.3:11.1)
0.77
0.55
<0.01
-34
(-63:6)
<0.05
V
̇O2max (ml·kg-1·min-1)
-0.024
(-0.04:0.03)
0.90
0.02
(-0.07:0.11)
0.64
-0.3
(-0.6:-0.04)
<0.05
V
̇Lamax (mmol·l-l·s-1)
-2.32
(-4.6:-0.02)
<0.05
-16.9
(-21.9:-11.8)
<0.01
3.6
(-0.11 :18)
0.63
Speed (m.s-1)
2.5
(0.84:4.1)
<0.01
5.1
(1.46:8.59)
<0.01
53
(42 : 63)
<0.01
Intervals (m)
0.0002
(-0.004:0.004)
0.92
0.0008
(-0.009 :0.01)
0.86
-0.10
(-0.13:-0.07)
<0.01
Recovery (s)
0.15
(-0.09:0.38)
0.21
0.31
(-0.21:0.83)
0.23
2.2
(0.6:3.8)
<0.01
The p-values associated with a single coefficient correspond to Student's t-tests whereas the p-values
associated with the overall model correspond to an ANOVA F-test.
Table 6. Simulated metabolic responses to 5 x 200 m training intervals with short, medium and long recovery
(start at 2:30, 2:40 and 2:50 min:s).
Mean V
̇O2
(ml·kg-1·min-1)
ΔV
̇O2
(1st-last)
% Mean
V
̇O2/
V
̇O2max
Mean
%V
̇La
(mmol·l-l·s-1)
ΔV
̇La
(1st-last)
%V
̇Lamax
Mean pH
ΔpH
Mean CPr
ΔPCr
%
Aerobic
% An.
lactic
% An.
alactic
5x200 m
@ 2:30
60.8 *
(54.4 - 67.2)
-1.1 *
(-1.5 -0.7)
0.8 *
(0.7 - 0.9)
1.0
(0.4 -1.7)
0.1 *
(0.0 - 0.1)
2.4 *
(0.4 – 4.4)
7.2
(6.8 - 7.6)
0.03
(-0.00 - 0.07)
8.2
(6.0 - 10.4)
0.6
(0.0 - 1.2)
88 *
(84 - 92)
8 *
(6 -11)
4 *
(2- 6)
5x200 m
@ 2:40
62.5*
(56.8 - 68.3)
-0.5
-0.8 -0.2)
0.8
(0.8 - 0.9)
1.4
(0.6 - 2.2)
0.1
(0.0- 0.2)
3.2 *
(0.9- 5.5)
7.0
(7.0 - 7.1)
0.05
(0.01 - 0.08)
7.9
(6.2 - 9.6)
0.7
(0.1 - 1.3)
85 *
(80 - 90)
10
(7 -12)
5 *
(3 - 7)
5x200 m
@ 2:50
63.9 *
(58.5 - 69.3)
-0.6 *
(-1.2 -0.1)
0.9 *
(0.8 - 0.9)
1.6 *
(0.8 - 2.4)
0.2 *
(-0.1 - 0.4)
3.6 *
(1.2 – 6.0)
7.5
(6.3 - 8.8)
0.07 *
(0.00 - 0.13)
7.0 *
(4.8 - 9.1)
0.9
(-0.2 - 1.9)
84 *
(80 - 89)
10
(8 - 12)
6 *
(3 - 8)
Data are mean (95% CI). Differences were identified using the Wilcoxon test for paired samples (* p <0.05).
Table 7. Simulated metabolic responses to 10 x 100 m training intervals with short, medium and long recovery
(start at 1:20, 1:30 and 1:45 min:s).
Mean V
̇O2
(ml·kg-1·min-1)
ΔV
̇O2
(1st-last)
% Mean
V
̇O2/
V
̇O2max
Mean
%V
̇La
(mmol·l-l·s-1)
ΔV
̇La
(1st-last)
%V
̇Lamax
Mean pH
ΔpH
Mean PCr
ΔPCr
%
Aerobic
% An.
lactic
% An.
alactic
10x100 m
@ 1:20
63.1
(57.46 - 68.68)
-3.3 #
(-4.6 -1.9)
0.8
(0.8 - 0.9)
1.8 *#
(0.9 - 2.7)
0.3 *
(-0.1 - 0.7)
4.0 *#
(1.4 - 6.7)
7.0
(7.0 - 7.0)
0.11 *
(0.02 - 0.21)
7.1
(5.1 - 9.1)
2.3 *#
(0.4 -
4.23)
77 *#
(70 - 84)
17 *#
(13 - 21)
6 *#
(3 - 8)
10x100 m
@ 1:30
63.1
(58.6 - 67.7)
-3.1 Ψ
(-5.2 - 1.0)
0.8
(0.8 - 0.9)
2.2 *Ψ
(1.1 - 3.3)
0.5
(0.1 - 0.8)
4.9 *Ψ
(2.0 – 7.9)
7.0
(7.0 - 7.0)
0.17 *
(0.05 - 0.28)
7.7
(5.8 - 9.7)
3.8 *
(0.5 - 7.0)
72 *Ψ
(65 - 80)
21 *Ψ
(16 - 26)
7 *
(4 - 9)
10x100 m
@ 1:45
62.6
(58.0 - 67.3)
-2.1 #Ψ
(-3.7 -0.5)
0.8
(0.8 - 0.9)
2.5 #Ψ
(1.5 - 3.4)
0.6 *
(0.2 - 1.0)
5.5 #Ψ
(2.7 - 8.3)
7.0
(7.0 - 7.0)
0.18
(0.07 - 0.29)
8.4
(6.6 - 10.3)
3.9 #
(0.9 - 6.9)
69 #Ψ
(60 - 77)
24 #Ψ
(18 - 31)
7 #
(5 - 10)
Data are mean (95% CI). Differences were identified using the Wilcoxon test for paired samples. Significant differences (p <0.05) were:
* 1:20 vs. 1:30; # 1:20 vs. 1:45; Ψ 1:30 vs. 1:45.
Table 8. Simulated metabolic responses to 20 x 50 m training intervals with short, medium and long recovery
(start at 0:45, 1:00 and 1:15 min:s).
50 m
Mean V
̇O2
(ml·kg-1·min-1)
ΔV
̇O2
(1st-last)
% Mean
V
̇O2/
V
̇O2max
Mean
%V
̇La
(mmol·l-l·s-1)
ΔV
̇La
(1st-last)
%V
̇Lamax
Mean pH
ΔpH
Mean CPr
ΔPCr
%
Aerobic
%
An. lactic
% An.
alactic
20x50 m
@ 0:45
60.4
(51.1 - 69.8)
-8.3 *#
(-12.9 -3.73)
0.8
(0.7 - 0.9)
1.7 #
(0.7 - 2.8)
-0.2 *
(-0.6 - 0.2)
3.8 *
(1.2 - 6.5)
7.0
(7.0 - 7.1)
0.13
(0.02 - 0.23)
9.0
(6.6 - 11.4)
4.1
(1.1 - 7.2)
60 *#
(59 - 71)
30 *#
(26 - 35)
5 *
(2 - 7)
20x50 m
@ 1:00
62.6
(58.0 - 67.26)
-3.6 *
(-6.4 -0.8)
0.8
(0.7 - 0.8)
2.4 Ψ
(1.7 - 3.2)
0.7 *
(-0.3 - 1.7)
5.2 Ψ
(3.4 - 7.1)
7.0
(7.0 - 7.1)
0.17
(0.05 - 0.29)
10.3
(7.5 - 13.1)
3.8
(1.3 - 6.2)
52 #Ψ
(44 - 60)
42 #Ψ
(36 - 49)
6
(4 - 7)
20x50 m
@ 1:15
62.6
(58.0 - 67.3)
-2.7 #
(-6.4 - 1.1)
0.8
(0.7 - 0.8)
2.7 #Ψ
(0.7 - 2.8)
0.7
(-0.6 - 2.0)
5.8 *Ψ
(4.4 - 7.3)
7.0
(7.0 - 7.1)
0.17
(0.06 - 0.29)
11,0
(9.1 - 12.9)
3.8
(1.7 - 5.9)
49 *Ψ
(40 - 58)
45 *Ψ
(37 - 54)
6 *
(5 - 7)
Data are mean (95% CI). Differences were identified using the Wilcoxon test for paired samples. Significant differences (p <0.05) were:
* 0:45 vs. 1:00; # 0:45 vs. 1:15; Ψ 1:00 vs. 1:15.
Table 9. Simulated metabolic responses to 40 x 25 m training intervals with short and long recovery (start at 30
and 45 s).
25 m
Mean V
̇O2
(ml·kg-1·min-1)
ΔV
̇O2
(1st-last)
% Mean
V
̇O2/
V
̇O2max
Mean
%V
̇La
(mmol·l-l·s-1)
ΔV
̇La
(1st-last)
%V
̇Lamax
Mean pH
ΔpH
Mean CPr
ΔPCr
%
Aerobic
%
An. lactic
%
An. alactic
40x25 m
@ 30 s
60.8 *
(54.1 - 67.5)
-14.4 *
(-18.0 -10.7)
0.8 *
(0.8 - 0.9)
2.2
(1.5 - 2.8)
-0.4
(-0.9 - 0.1)
4.7
(2.9 - 6.5)
7.0
(7.0 - 7.1)
0.20
(0.09 - 0.31)
9.5
(8.0 - 10.9)
7.3 *
(3.6 – 11.0)
46 *
(43 - 50)
50 *
(47- 53)
4
(3 - 5)
20x50 m
@ 45 s
54.6 *
(50.6 - 58.6)
-7.6 *
(-11.6 -3.6)
0.7 *
(0.7 - 0.8)
2.2
(1.8 - 2.6)
-0.3
(-1.3 - 0.7)
4.8
(3.8 - 5.8)
7.1
(7.0 - 7.1)
0.11
(0.02 - 0.19)
12.0
(9.1 - 15.0)
3.7 *
(2.4 - 5.1)
37 *
(34 - 40)
59 *
(57 - 62)
4
(3 - 5)
Data are mean (95% CI). Differences were identified using the Wilcoxon test for paired samples. Significant differences (p <0.05) were:
* 30 vs. 45 s
