Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

Sprinters motor signature does not

change with fatigue

Mohamed-Amine Choukou, Guillaume


Laffaye & Anne-Marie Heugas-De
Panafieu

European Journal of Applied


Physiology

ISSN 1439-6319

Eur J Appl Physiol


DOI 10.1007/s00421-011-2107-9

1 23
Your article is protected by copyright and
all rights are held exclusively by Springer-
Verlag. This e-offprint is for personal use only
and shall not be self-archived in electronic
repositories. If you wish to self-archive your
work, please use the accepted authors
version for posting to your own website or
your institutions repository. You may further
deposit the accepted authors version on a
funders repository at a funders request,
provided it is not made publicly available until
12 months after publication.

1 23
Author's personal copy
Eur J Appl Physiol
DOI 10.1007/s00421-011-2107-9

ORIGINAL ARTICLE

Sprinters motor signature does not change with fatigue


Mohamed-Amine Choukou Guillaume Laffaye

Anne-Marie Heugas-De Panafieu

Received: 11 April 2011 / Accepted: 28 July 2011


Springer-Verlag 2011

Abstract The aim of this study was to investigate human PCA revealed three different sprint profiles explaining 88.2%
adaptations to fatigue induced by track sprint repetitions. of the total variance: the contact-time-pattern (39.46%),
Eight male sprinters were asked to run 4 9 100 m as quickly force-pattern (27.96%) and stride-pattern (20.77%). Two
as possible with 3 min of recovery. Subjects were filmed different motor signatures were identified with fatigue. In the
(50 Hz) in order to measure stride length and frequency. first, athletes switch from the key variable to another when
Velocity was measured by means of radar (250 Hz) while exhausted without changing their motor behavior (during
contact and flight times were registered wirelessly by two Repetition 3 and/or Repetition 4). In the second, athletes do
pressure sensors (400 Hz) embedded under the insole of the not change their motor behavior with fatigue.
subjects shoes. Contact and flight times were used to calcu-
late stiffness. In addition, blood samples were taken prior to Keywords Spring mass model  100 m  Exhaustion 
warm-up, 1 min after each 100-m sprint and every 2 min after Running  Field conditions
the last repetition until a lactate peak ([BLa]) was reached.
[BLa] did not affect mechanical and stride parameters. Inter- List of symbols
series ANOVA showed that velocity decreased significantly
Mechanical parameters
(-3.55%) between Repetition 1 (8.18 0.29 m s-1) and
CT Ground contact time (ms)
Repetition 4 (7.89 0.42 m s-1), while [BLa] increased
FT Flight time (ms)
from 6.74 1.15 to 13.58 1.48 mmol l-1 (p \ 0.05). The
V Velocity of the centre of mass (m s-1)
first main result was that leg stiffness remained constant until
Fmax Maximum leg force (kN)
Repetition 3 and then dramatically increased at Repetition 4,
DCoM Vertical displacement of the centre of mass (m)
whereas vertical stiffness remained constant throughout all
L Leg length at standing position
four repetitions. This behavior could be considered to be a
DL Shortening of the leg (m)
neuromuscular adaptation to fatigue used by skilled athletes.
kleg Leg stiffness (kN m-1)
The second main result was that velocity decreased during the
kvert Vertical stiffness (kN m-1)
second phase (3080 m) of the entire 100 m. In addition, a
CoM Body centre of mass

Communicated by Toshio Moritani. Stride parameters


SF Stride frequency (Hz)
Electronic supplementary material The online version of this
article (doi:10.1007/s00421-011-2107-9) contains supplementary SL Stride length (m)
material, which is available to authorized users.
Metabolic parameter
M.-A. Choukou (&)  G. Laffaye  [BLa] Blood lactate rate (mmol l-1)
A.-M. Heugas-De Panafieu
UR CIAMS, Motor Control and Perception Group, Other abbreviations
Sport Sciences Department, Universite de Paris Sud,
Batiment 335, Bureau 38, 91405 Orsay Cedex, France SMM Spring mass model
e-mail: mohamed-amine.choukou@u-psud.fr SSC Stretch shortening cycle

123
Author's personal copy
Eur J Appl Physiol

CNS Central nervous system constitutes the spring-mass model (SMM). In the literature,
PCA Principal component analysis the SMM is commonly used to assess both leg (kleg in
kN m-1) and vertical stiffness (kvert in kN m-1). The latter
makes it possible to model the vertical motions of the CoM
during the stance phase (Blickhan 1989; McMahon and Cheng
Introduction 1990; Alexander 1992; Farley and Gonzalez 1996). However,
there are some discrepancies in the literature about the pos-
Although fatigue during the 100-m sprint has been widely sible link between performance and leg stiffness. The latter
investigated from a physiological point of view (Hirvonen has been shown to be correlated with 100-m performance
et al. 1987; Nummela et al. 1992; Hautier et al. 1994; Bret during official national events (Bret et al. 2001). On the con-
et al. 2001), its mechanical causes are incompletely under- trary, the mechanical parameters of novices, including leg
stood. The 100-m sprint requires maximum energy produc- stiffness, remain constant throughout the 100 m regardless of
tion in just a few seconds. Metabolic energy is provided velocity (Morin and Belli 2003; Morin et al. 2006).
mainly by anaerobic glycolysis and the phosphocreatine From a metabolic point of view, there are some plausible
metabolisms in skeletal muscle cells because of inadequate explanations for fatigue with the decrease in ATP, phospho-
oxygen supply to the mitochondria. Anaerobic glycolysis has creatine and glycogen, and a decrease in blood pH due to an
been reported to contribute greatly to metabolic energy increase of protons H? (Cheetham et al. 1986). This was
production during the running of a 100-m sprint: 55% challenged by Pedersen et al. (2004), who conducted experi-
(Hautier et al. 1994) 57% (Hirvonen et al. 1987), and 76.2% ments on isolated muscles and suggested that acidosis has little
(Bret et al. 2001). The latter can be assessed in field condi- detrimental effect, or may even improve muscle performance
tions by measuring post-exercise blood lactate rates. Some during high-intensity exercise. To reconcile these findings, it is
investigators have noted that [BLa] is correlated to 100-m hypothesized that severe plasma acidosis in humans might
performance (Bret et al. 2001) unlike previous researchers impair exercise performance by causing reduced nervous flow
for whom the two parameters are independent (Nummela to the muscles (Gibson and Noakes 2003).
et al. 1992; Hautier et al. 1994). This proves that energy yield The biomechanical explanations for fatigue during run-
varies according to the training level. The latter reflects ning also remain unclear. There are actually few investiga-
muscular efficiency, i.e. metabolic energy use and elastic tions on the effect of fatigue on mechanical stiffness during
energy restitution during the stretch shortening cycle (SSC). the 100-m sprint in real conditions. For instance, Bret et al.
This is why it would be useful to divide the 100 m into parts (2002) demonstrated the contribution of leg stiffness in
in order to better understand whether metabolic energy maintaining speed over 100 m, but the way it was regulated
depletion affects mechanical behavior during running. throughout the sprint was unknown. To our knowledge,
The 100 m is commonly divided into three parts: Morin et al. (2006) are the only researchers to study the
acceleration, speed maintenance and deceleration. The first change in SMM characteristics with fatigue induced by
phase requires high mechanical power so as to overcome sprint repetition in novices. Their main results show that leg
inertia during the first steps and to increase forward stiffness remains constant throughout the 100 m and
acceleration until approximately 30 m (Mero et al. 1992; throughout repetitions. Fatigue did not influence kleg
Chelly and Denis 2001). Sprinters maintain maximal speed directly, but altered some of the stride and SMM parameters
up to 60 m by storing and restituting elastic energy within such as step frequency, which decreased, and the vertical
the musculo-tendinous structure of their lower limbs, motions of the CoM, which increased.
respectively, during the eccentric and concentric phases of The purposes of this study were (1) to identify whether
the SSC (Blickhan 1989; McMahon et al. 1987; McMahon blood lactate concentrations affect sprint performance, (2)
and Cheng 1990). However, anaerobic glycolysis is unable to determine leg stiffness changes with sprint repetition and
to produce great energy thereafter. This is why a deceler- in the 100-m phases in field conditions in skilled sprinters,
ation phase is observed at the end of the 100 m. During this and (3) to determine sprint motor pattern behavior as a
phase, stride frequency decreases conversely to stride function of fatigue induced by four all-out 100-m sprints.
length, which increases (Mero et al. 1992). It has been
reported that the depletion of speed in experts was linked to
the decrease in vertical stiffness during the 100-m sprint, Methods
but no mechanical parameters were linked to performance
during this phase (Morin and Belli 2003). Participants
Running has been considered a bouncing gait in which the
lower limbs behave as a linear spring during each stance Eight skilled male sprinters, competing at the regional
phase. Thus, the leg-spring loaded by the entire body mass level of the French track-and-field championships

123
Author's personal copy
Eur J Appl Physiol

(11.54 0.37 s in the 100 m), exhibiting no injuries to the installed 3 m behind the runners and calibrated at a fre-
lower limbs, participated in this study. Their physical char- quency of 250 Hz. Raw speed data were filtered (by a four-
acteristics were age 19 2 years, body mass 72.88 5.78 order zero-lag Butterworth filter) using ATS SystemTM
kg, height 1.80 0.07 m and leg length 0.93 0.04 m. acquisition software. This device was placed on a tripod 3 m
Subjects were instructed to continue their normal diets behind the sprinter at a height of 1 m, which corresponded
throughout the testing procedures and were advised to refrain approximately to the mean height of the CoM. Contact and
from caffeine and alcohol during the day preceding testing. flight times were measured using two pressure sensors (fre-
Each volunteer signed a written informed consent statement quency 250 Hz) embedded under the insoles of the shoes
after receiving verbal and written descriptions of the proce- (Locotest, Techno-concept, France) connected to modules
dures which had been approved by the Ethics Committee of (right ? left) that were tied to the top of the corresponding
the University of Paris-Sud and was informed of the risks and shoe. Each sensor was affixed between the insole and the
benefits of participating in this study. middle of the forefoot. The entire device was attached to the
participant and connected to the experimenters laptop via
Bluetooth (see Fig. 1). Figure 1 illustrates the locations of
Protocol
the devices and protocol organization.
A variety of devices were used to measure comple-
After performing a standardized warm-up, participants were
mentary stride parameters (frequency and amplitude). The
instructed to run as quickly as possible for four repetitions of a
trials were taped (SONY, Sony Electronics Inc., Paris,
100-m sprint (R1, R2, R3 and R4) on a tartan track.
France) at a frequency of 50 Hz. The camcorder was
The repetitions were separated by 3-min recovery periods
positioned 7 m beside the middle of the 100-m lane. Stride
during which the runners returned to the start of the sprint
length (SL in m), which is useful for calculating step fre-
course. The experiment was conducted on an outdoor track
quency (SF in Hz), was measured only by video analyses.
from a standing position so that the sprinters were in a real
SL corresponded to the distance separating the initial
context similar to their training sessions. The athletes wore
ground contact point of a given foot up to the contact point
their own spiked shoes, their usual uniform, and the devices
of the opposite foot. SF was calculated as the opposite of
described below. They confirmed that none of acquisition
the time of a running cycle, i.e. the sum of contact and
systems disturbed their movements or sensations.
flight times (Eq. 1):
1 1
Materials and data collection SF 1
T CT + FT
Instantaneous forward velocity (V) was measured by means In addition, blood lactate concentrations were measured
of radar (Stalker ATS, Applied Concepts, Inc, Texas, USA) by means of a portable lactate analyzer (Lactate ProTM).

Fig. 1 Experimental protocol

4 * 100 m sprint (3 mn of recovery)

Instruction : run as fast as possible

30m 60m 80m


Blood Lactate
Portable

analyser

first phase second phase third phase

123
Author's personal copy
Eur J Appl Physiol

Blood samples were taken at rest ([BLa0]), 1 min after each met. Tests of normality of distribution and skewness
100 m ([BLax]) during the recovery period and every (ShapiroWilk test) revealed no abnormal data pattern with
2 min after R4 until a peak of [BLa] was reached. The p-values ranging between 0.07 and 0.97. ANOVA with
[BLa0] was considered a reference value so that blood Fishers post-hoc test were performed for the four repeti-
lactate rate could be estimated as the difference between tions and the three 100-m phases. This allowed us to
each [BLax] and the rest value (Eq. 2): observe whether the SMM properties and stride parameters
DBLa BLax   BLa0  2 were altered within each 100-m sprint and throughout the
repetitions. In addition, a Pearsons correlation test was
used to determine whether correlations between blood
Leg and vertical stiffness calculation method lactate, performance, and mechanical parameters of the
four repetitions exist. The statistical significance level was
Leg and vertical stiffness were calculated using the sine-wave set at p \ 0.05. Additionally, PCA was performed on the
method elaborated by Morin et al. (2005) (Eq. 3). According data obtained from the 32 sprints (8 subjects 9 4 sprints)
to this method, the F(t) curve can be modeled as a sinus in order to identify the principal components summarizing
function Y t A  sin xt U, where Fmax represents the nine independent variables (SL, SF, CT, FT, Fmax,
the amplitude of the wave (A in the function), (P/CT) the DCoM, DL, kleg and kvert). The PCA was obtained from the
angular frequency (x), and the phase U is equal to zero. STATISTICA package (version 5.5) using the procedure
In this study, we considered that CoM movements were described by Kollias et al. (2001). The number of principal
symmetric between 30 and 100 m. Thus, the sine-wave components in the pattern matrix extracted by the PCA was
method could be applied over this distance, considering the chosen with an eigenvalue greater than 1. The original
velocity loss at the end of 100 m as negligible: matrix was rotated to extract the appropriate variables,
  using a normalized VARIMAX rotation (orthogonal rota-
P
F t Fmax  sin t 3 tion). This rotation allowed an earlier labeling of the
CT
principal components. In order to characterize running
Considering that maximal vertical force (Fmax) reaches profiles, individual repetitions were plotted in two planes
its maximal value when the CoM is low (McMahon and containing the three principal components. Figure 2 illus-
Cheng 1990), kvert was calculated (Eq. 4) as the ratio of trates a first plane with the first (first PC) and second
Fmax (Eq. 5) to the maximal downward displacement of the (second PC) components, while Fig. 3 shows a second
CoM during contact (DCoM in m) (Eq. 6) (Farley and plane with the first PC and the third (third PC) component
Gonzalez 1996): (see Figs. 2, 3).
kvert Fmax  D1
CoM 4
with Results
 
P FT
Fmax m  g   1 5 Results of each parameter were compared in the three
2 CT
100-m phases and in the four repetitions. In accordance
Fmax CT2 CT2 with the literature, each 100 m was divided into three parts:
DCoM   2 g 6
m P 8 030, 3080 and 80100 m (Delecluse et al. 1995; Chelly
and Denis 2001; Morin and Belli 2003).
Leg stiffness was calculated (Eq. 7) as the ratio of Fmax
to leg-spring peak displacement (DL) during contact (Eq. 8)
(McMahon and Cheng 1990): Relationship between blood lactates and sprint
kleg Fmax  DL 1
7 performance

with The subjects ran R1 in 12.43 0.55 s, representing 92.84%


v
u
u  2 ! of their personal best over 100 m. The velocity decreased
t V  CT significantly [F(3, 21) = 4.03; p \ 0.05] by 3.55% from
DL L  L2  DCoM 8
2 R1 (8.18 0.29 m s-1) to R4 (7.89 0.42 m s-1). As
hypothesized, blood lactate concentrations increased sig-
Statistical analysis nificantly from 6.74 1.15 mmol l-1 1 min after R1, to
13.58 1.48 mmol l-1 1 min after R4 [F(3, 21) = 35.37;
All descriptive statistics were used to verify whether the p \ 0.05]. A peak of 14.76 mmol l-1 was reached 2 min
basic assumption of normality of all studied variables was after R4. The blood lactate rate was correlated (r = 0.83;

123
Author's personal copy
Eur J Appl Physiol

stiffness increased significantly by 19.68% with repetition


[F(3, 21) = 4.27; p \ 0.05], whereas kvert remained con-
stant. The post-hoc test revealed that leg stiffness remained
constant between R1 (20.92 4.66 kN m-1) and R3
(19.56 3.1 kN m-1), and increased dramatically at R4
(25.03 2.21 kN m-1). The imposed exhaustion did not
alter vertical stiffness [F(3, 21) \ 1], which varied from
74.64 7.7 kN m-1 during R1 to 72.12 5 kN m-1
during R4, representing only 3.38% of the decrease.

Stride parameters and fatigue

Mean contact time was 125 s and mean flight time was 169 s.
Both parameters remained constant throughout the four
Fig. 2 Individual scores for each subject and each repetition of repetitions [F(3, 21) \ 1]. Vertical leg displacement also
sprint on the 1st and 2nd principal components. The first principal remained constant [F(3, 21) \ 1] in spite of fatigue
component represents the individual scores on the CT-component and (17.24 2 cm), which was not the case for stride frequency
the second principal component represents the individual scores on (3.89 0.25 Hz) and stride length (2.07 0.13 m). SF did
the force-component. Each sprinter corresponds to a symbol associ-
ated with a letter and numbered from 1 to 4. The number corresponds not change significantly between R1 (3.99 0.25 Hz) and
to the row of the sprint repetition R4 (3.90 0.26 Hz) (-2.07%) [F(3, 21) \ 1]. Similarly,
stride length did not decrease significantly (-1.63%)
between R1 (2.06 0.14 m) and R4 (2.03 0.12 m) [F(3,
21) \ 1] (Table 1).

SMM behavior during the three phases of the 100 m

None of the SMM properties varied during the acceleration


phase, i.e. between 0 and 30 m [F(3, 21) \ 1]. The only
change observed during the second phase (3080 m) of all
sprints concerned velocity, which decreased significantly
[F(3, 21) = 6.53; p \ 0.05] from 9.06 m s-1 during R1 to
8.60 m s-1 during R4. The change of velocity was corre-
lated with stride length during the second phase (r = 0.80;
p \ 0.05). However, velocity stopped decreasing at the end
of the sprints (8.438.81 m s-1), i.e. between 80 and
100 m. During this deceleration phase, none of the stride
Fig. 3 Individual scores for each subject and each repetition of sprint parameters (SF, SL, CT, FT, DL, DCoM) was altered with
on the 1st and 2nd principal components. The first principal repetition [F(3, 21) \ 1]. Leg and vertical stiffness also
component represents the individual scores on the CT-component
remained constant [F(3, 21) \ 1].
and the third principal component represents the individual scores on
the stride-component. Each sprinter corresponds to a symbol associ-
ated with a letter and numbered from 1 to 4. The number corresponds Principal component analysis
to the row of the sprint repetition
The PCA revealed three principal components accounting
for 88.2% of total variance and including all the studied
p \ 0.05) to velocity only during R3 (12.04 mmol l-1) variables (Table 2).
(Table 1). The first PC (Table 2), which was rotated, accounted for
39.46% of variance with an eigenvalue of 3.64. As shown
Changes in mechanical parameters under the fatigue in Table 2, this CT-component was highly loaded with the
condition following parameters: negative values of kleg (-0.861) and
kvert (-0.852) explained that they are negatively related to
Force averaged 2.68 0.26 kN and did not vary signifi- DL (0.963) and CT (0.867). This first PC assumes that an
cantly with sprint repetition [F(3, 21) \ 1]. Mean kleg increase in leg and vertical stiffness is obviously linked to a
was 21.28 3.62 kN m-1. Contrary to expectations, leg decrease in the CT and leg lowering.

123
Author's personal copy
Eur J Appl Physiol

Table 1 Blood lactate rates, stride and mechanical parameters [mean (SD)] during the four repetitions of 100 m
R1 R2 R3 R4 ANOVA % change

V100 (m s-1) 8.18 (0.29) 8.12 (0.27) 8.03 (0.32) 7.89 (0.42) (14, 24)* -3.55
SL (m) 2.06 (0.14) 2.13 (0.12) 2.09 (0.15) 2.03 (0.12) (12, 24)* -1.63
SF (Hz) 3.99 (0.25) 3.83 (0.23) 3.87 (0.29) 3.90 (0.26) (12, 13)* -2.07
CT (ms) 122.91 (5.7) 126.96 (6.4) 129.02 (5.1) 125.03 (3.7) ns 1.72
FT (ms) 168.5 (12.2) 171.61 (9.8) 163.9 (17) 174.53 (10.5) ns 3.58
Fmax (kN) 2.70 (0.22) 2.69 (0.28) 2.59 (0.28) 2.73 (0.32) ns 1.07
DCoM (cm) 3.74 (0.3) 3.94 (0.2) 3.86 (0.4) 3.93 (0.2) ns 5.06
DL (cm) 17.21 (2.4) 17.91 (2.4) 17.9 (2.2) 15.9 (0.9) ns -7.26
Kleg (kN m-1) 20.92 (4.66) 19.64 (4.53) 19.56 (3.10) 25.03 (2.21) (14, 24, 34)* 19.64
-1
Kvert (kN m ) 74.64 (7.7) 69.26 (7.35) 68.61 (7.5) 72.12 (5) ns -3.38
[BLa] (mmol l-1) 6.74 (1.15) 10.60 (1.82) 12.04 (2.24) 13.58 (1.48) (12, 13, 14, 24, 34)* 101.48
(% change) corresponds to the percentage of variation of parameters between R1 and R4 for all subjects
ns non-significant
* (serie-serie) significantly different from the other repetitions (p \ 0.05) with Fishers post hoc test

Table 2 Results of the PCA


Variables Factor loadings Commonalities
showing for each variable factor
loading and commonalities as F1 F2 F3
well as eigenvalue and
percentage of variance for each SL -0.856 0.905
rotated principal component SF 0.966 0.908
CT 0.867 0.800
DL 0.963 0.868
kleg -0.861 0.705
kvert -0.852 0.808
FT -0.955 0.737
Dyc 0.840 0.878
Fmax -0.778 0.847
Eigenvalue 3.64 3.27 1.09
% variance 39.46 27.96 20.77

The second PC (see Table 2) explained 27.96% of the positive. Then, we used a step-by-step multiple regression
variance and its eigenvalue was about 3.27. This compo- analysis (Foucart 1999; Dompnier et al. 2007), entering the
nent was positively loaded by DCoM (0.840) and negatively individual factor scores on the three components as inde-
loaded by Fmax (-0.778) and FT (-0.955) as shown in pendent variables and mean velocity as dependent vari-
Table 2. The Force-component showed a relationship ables. In order to understand the effect of fatigue, we
between force production and stride characteristics. The divided the data into two groups: the first includes the
decrease in CoM displacement is linked to the decrease in velocity of R1 and R2, and the second the velocity of R3
Fmax and in FT. and R4. The equations found for series 1 (R1R2) and 2
The third PC (see Table 2) explained 20.77% of the (R3R4) are reported in (Eq. 9) and (Eq. 10), respectively
variance with an eigenvalue of 1.09. It was highly loaded (see Eqs. 9, 10):
by SF (0.966) and SL (-0.856). This stride-component
showed a link between stride length and frequency. This VR1R2 0:22 F1 0:422 F2 0:031 F3
 
means that athletes who increase stride frequency simul- 8:1 with r 2 0:26; p 0:3 9
taneously decrease stride length. VR3R4 0:06 F1 0:099 F2 0:590 F3
In order to understand the link between components, the  
7:8 with r 2 0:37; p 0:2 10
factor scores of each participant for the three dimensions
were recorded and used as independent variables after data These equations allowed us first to understand the
normalization to change negative factor loadings to connection between the three components. The contact time

123
Author's personal copy
Eur J Appl Physiol

component (F1) is negatively linked to the other two, mechanical parameters of sprint. The results of this study
explaining that the force and stride components increase showed that [BLa] increased gradually and significantly
when the contact time component decreases. Second, the (p \ 0.05) from 6.74 mmol l-1 after R1 to 13.58 mmol l-1
relative contribution of each component in both equations for after R4. [BLa] values are close to values reported during a
R1 and R4 is different. Only the signs of the three components 100-m sprint of a national competition (12.5 mmol l-1
do not change, showing the same relationship between (Bret et al. 2001), and 8.5 mmol l-1 (Hautier et al. 1994)),
components independently of fatigue. Without fatigue, the except after R1. This could be due to the shortness of the
great contributor is the force pattern (56%), which then time (1 min) between blood sampling and the end of R1.
switches to stride pattern with fatigue, which contributes 50% This time period was probably insufficient to allow lactic
of total variability. This shows that a high level of vertical acid to be buffered and transported from the skeletal
force is a major factor of velocity, confirming the recent study muscles to the blood.
of Morin et al. (2011). With fatigue, the sprinter is unable to Blood lactate rates affected neither the 100-m mechan-
produce a high level of force but is better able to regulate the ical parameters (CT, FT, Fmax, DCoM, DL, kleg, kvert), nor the
stride component, i.e. stride length and stride frequency. stride parameters (SL, SF), and thus cannot explain the
Plotting the mean individual scores on the three PCs, as causes of the decrease in velocity with repetition. Hirvonen
shown in Figs. 2 and 3, we understand the effect of et al. (1987) reported that performance in short supra-
exhaustion on different individual patterns (see Figs. 2, 3). maximal exercise depends primarily on the ability to use
The horizontal axis of Fig. 2 corresponds to the first PC, high-energy phosphate stores, and secondarily most of the
namely the CT component. Sprinting performance with a energy must then be produced by glycolysis. So, the
positive high level on the first PC would demonstrate high decline in running velocity is not related to metabolic
values for CT, DL, and low values for kleg and kvert. This acidosis via lactate accumulation, but could rather be
pattern was employed by athletes A and G oriented linked to the availability of high-energy phosphate. The
to stiffness with negatives component scores ranging from 3 min of passive recoveries used in this study allowed
-2.97 to -0.79 and C and D who relied rather on incomplete recoveries between successive sprints creating
contact time and leg lowering reducing with positive scores a fatigue state and hence a decrease in performance
(0.143.94) on the first PC. (Bogdanis et al. 1995; McCartney et al. 1986). Although
The vertical axis of Fig. 2 indicates the second PC, which trained sprinters are able to deplete and resynthesize high-
was characterized as the force-component. A positive high energy phosphate stores more effectively than untrained
level on this second PC indicates a high value of DCoM and sprint participants, the recovery periods were likely inad-
low values of strength and flight time. This pattern consists in equate for creatine phosphate to recover fully and hence to
strength use in order to increase flight time and decrease maintain a high level of velocity (Bogdanis et al. 1995;
CoM vertical displacement. The component scores ranged Rehunen et al. 1982). Furthermore, the decrease in creatine
from -0.7 to -3.7 corresponding to subjects F and H. phosphate availability combined with the continued
Sprinters B and E showed neither first nor second PC, attempt to generate maximum power should stimulate
i.e. a neutral pattern consisting of a third PC, namely the glycolysis and lead to the increase of [BLa] throughout the
stride-component represented by the vertical axis of Fig. 3 four 100-m sprints.
with component scores ranging from -0.24 to 1.58 (see As suggested by Cairns (2006), lactic acid, which is
Fig. 3). Running with a positive value on this component considered a major cause of skeletal muscle fatigue (i.e.
would show high values for SF and low values for SL. decline of muscle force or power output leading to
impaired exercise performance), is questionable especially
if accompanied by a high capacity for lactate removal as is
Discussion expected in trained sprinters. Other ions such as pyruvate,
phosphate and citrate ions could cause post-exercise aci-
The main purposes of this study were to determine (1) the effect dosis (Cairns 2006). In the case of runners who rely on
of metabolic fatigue on sprint mechanical parameters, (2) the Fmax as a regulator of flight time or of CoM vertical dis-
role of lower limb stiffness in maintaining speed, and (3) the placement, we could deduce that an increase in lactate
behavior of mechanical and stride parameters throughout sprint concentrations related to high levels of metabolic acidosis
repetitions and in the three phases of the 100 m. would alter mechanical output by decreasing muscular
contraction rates (Nummela et al. 1992). Our unexpected
Blood lactates and sprint repetition findings contradicted this hypothesis since there was no
significant decrease in Fmax even though velocities
The first aim of this study was to identify possible rela- decreased. So, the depletion of phosphocreatine stores
tionships between blood lactate concentrations and cannot be considered as the sole determinant of muscle

123
Author's personal copy
Eur J Appl Physiol

fatigue. The mechanism involved in force production could conditions (17.119.5 kN m-1) (Morin et al. 2006). This
be another (Boyas and Guevel 2011). In addition, other difference can be explained by higher values of mean Fmax
studies showed that causes of fatigue are more likely to be in skilled sprinters (2.67 kN) compared with novices
neuromuscular (Pedersen et al. 2004; Cairns 2006). Indeed, (2.28 kN) (Morin et al. 2006). The high level of leg stiff-
different cases of adaptations to fatigue were observed in ness observed in this study agrees with the findings of
our participants. The CNS seems to be constrained to adapt Farley and Gonzalez (1996) and Morin et al. (2007). The
the running pattern during each repetition. One of our authors considered leg stiffness as a major determinant of
major results is that the three running profiles remained the SSC behavior. Indeed, in our study, CT, DCoM and kvert
same during the four 100 m. That will be discussed further values declined dramatically with fatigue in novices
in the last chapter. (Morin et al. 2007), whereas these parameters remained
unchanged for skilled athletes. For many authors (Farley
The major role of leg stiffness in maintaining speed and Gonzalez 1996; Morin et al. 2007; Laffaye et al. 2005),
shortening CT values or increasing SF values are a good
Our second purpose was to determine changes in lower way to maintain high stiffness values. This study clearly
limb stiffness with repetitions of 100 m at maximum speed. shows that stride frequency decreased until R3 whereas
Velocity decreased by 3.55% between R1 (8.18 m s-1) and CT, FT, Fmax, DCoM, DL and kvert remained unchanged,
R4 (7.89 m s-1) (Table 1). These results agree with the making it possible to maintain high stiffness values. This
literature for R1 (Bret et al. 2001; Morin et al. 2006) and was not the case in novices, whose CT, DCoM and kvert were
correspond to performances of trained athletes. The affected greatly by fatigue (Morin et al. 2006).
decrease in velocity of our participants was lower than the
decrease in velocity observed in novices (11.6%) during Does fatigue alter the sprinters motor signature?
similar conditions (Morin et al. 2006). This difference
could be explained by training status. In other words, The third aim of this study was to determine the organi-
skilled sprinters get tired less easily than novices during zation of mechanical and stride parameters throughout
such exercise. The literature regarding the link between leg sprint repetitions and in the phases of the 100 m.
stiffness and running sprint velocity is still contradictory. A main result of this study showed that no change was
For instance, Arampatzis et al. (1999) demonstrated observed during the first and the last phase of the 100 m as
that leg-spring stiffness influences running velocity dur- hypothesized (p \ 0.05). The only significant result was a
ing running over a force plate at different speeds decrease in velocity at the end of the second phase
(2.56.5 m s-1). This was not the case for Farley et al. (-5.07%) (p \ 0.05). The first part of the sprint is more
(1993), who indicated that leg stiffness was almost constant likely to be an important phase since it requires high leg
with increasing running velocity in animals. The explana- power values, i.e. high force and high acceleration, which
tion of sprint performance with leg stiffness as a major are both responsible for stride length and frequency
parameter of the SMM seems incomplete. This is why an increase (Chelly and Denis 2001). Nevertheless, the second
observation of the variation of both of the other SMM phase can be considered as the major part of the 100-m
parameters and stride parameters is required. sprint since the athlete has to maintain the velocity
The participants of this study showed no alteration achieved during the first phase. The results of this study
(-2.07%) in stride frequency between R1 and R4, compared showed that the decrease in velocity with fatigue occurred
with novices (-8.03%) (Morin et al. 2006). It has been during the second phase of the 100 m.
shown (Morin et al. 2007) that stride frequency indirectly In this study, there was no change in speed during the
affects SMM properties through its effect on contact time. first phase of the four sprints. This means that parameters
Moreover, Farley and Gonzalez (1996) suggested that producing acceleration are not affected by fatigue induced
performing high step frequencies requires a high leg stiff- by sprint repetition because the sprinter is using more
ness level. In this study, the participants maintained their power in order to lengthen the steps. In other words, the
contact time almost constant (1.72%) between R1 and R4, contractions are longer because the ground contact time
while the novices contact time increased significantly duration is very high between 0 and 30 m. A long CT
(?14.7%) (Morin et al. 2006). Based on these consider- seems to be less sensitive to fatigue during this first phase,
ations, the skilled sprinters were able to maintain nearly the contrary to the second phase, which requires shorter CT
same CT throughout the repetitions by minimizing the and higher SF to be able to run quickly (Nummela et al.
decrease in SF. 1996). This means that the second phase of the 100 m
The kleg values ranged from 20.92 kN m-1 during R1 to requires a high neuromuscular yield. During the decelera-
25.03 kN m-1 during R4. Leg stiffness values were higher tion phase, the velocity did not change in the four 100-m
than that of the results of previous studies in similar repetitions. Athletes CT did not increase at the end of the

123
Author's personal copy
Eur J Appl Physiol

sprint, which could explain the maintenance of a high level the sum of joint stiffnesses (i.e. ankle, knee and hip stiff-
of speed in accordance with Nummela et al. (1996), who ness). In other words, joint stiffness is required for landing
showed a strong link between velocity and CT during the while leg stiffness allows optimum use of the SSC. Thus,
400-m sprint. So, maintaining short CT seems to be an the CNS could guide the runners lower limb as a whole
efficient motor behavior in order to avoid a decline in segment, when exhausted, instead of governing three dif-
velocity at the end of the 100 m. This strategy is valid in ferent segments (thigh, leg and foot) and their corre-
skilled sprinters since CT and velocity remained constant sponding joints. So, the leg properties can be controlled by
in the third phase of the 100 m. regulating its stiffness, as suggested by Forestier and
Except for the velocity decrease during the second Nougier (1998) and Rabita et al. (2008). Forestier and
phase, we did not notice any changes in either stride or in Nougier (1998) reported increased movement rigidity
SMM parameters in the three phases. This constant state with fatigue which involved a new motor coordination. The
could be considered as a reorganization of stride parame- authors concluded that increasing the stiffness of the upper-
ters in order to adapt to muscular fatigue, namely the body multi-joint system would simplify movement execu-
decrease in Fmax. With fatigue, the lower limb attempted to tion and control with fatigue. These findings could explain
decrease CoM vertical displacement so as to optimize the unexpected increase in leg stiffness between R3 and R4
vertical stiffness, which remained unchanged with fatigue explained above. In our study, the neuromuscular regula-
while leg stiffness changed only between R3 and R4. These tions of skilled sprinters were probably designed to adapt
results are in accordance with the findings of Gibson and leg stiffness according to the fatigue state. However, the
Noakes (2003), showing such an adaptive phenomenon to way of running cannot be exclusively explained by leg
fatigue. Thus, leg stiffness could constitute a controlling stiffness analysis. That is why possible combinations
variable (Gibson and Noakes 2003) of a pacing strategy, between SMM and stride parameters during a 100-m sprint
used by the CNS to reduce the effects of fatigue. We and throughout repetitions were investigated.
suppose that the pacing strategies used by our participants The model found by the PCA accounted for 88.2% of
are quite efficient since velocity stopped decreasing during the total variance and the nine studied variables are sum-
the third phase of the four 100-m repetitions. marized in three components. This could explain individual
One of the main results of this study was that leg stiff- behaviors. Indeed, individual plotting of this PCA reveals
ness remained constant until R3 and then, surprisingly, three different patterns which capture all the individual
increased and reached its highest value during R4 signatures of the athletes motor behavior during sprint
(?19.68% of R1) (Table 1). This leg stiffness change repetition, independently of their running velocities.
contrasts with Morin et al. 2006) on the same protocol of
sprint repetition in novices, with Hobara et al. (2010) on The contact-time component
the 400-m sprint, and with Rabita et al. (2011) on long-
distance running. In these studies, the decrease in leg The first observed pattern was on the first PC of the PCA,
stiffness is linked to a decrease in either vertical force or which linked CT and DL throughout sprint repetition so as to
stride frequency, or in a maximal downward shift of the optimize leg and vertical stiffness. Athlete G presented
CoM. However, in our study, neither delta DCoM nor Fmax negative scores for this component over all repetitions. This
decreased between repetitions in our participants. By tak- means that G relied on high levels of leg (mean value:
ing into account these results, it appears that the onset of 17.4 kN m-1) and vertical stiffness (mean value:
fatigue was not mainly metabolic but rather the result of an 73.68 kN m-1), which could contribute to resistance to
interaction between metabolic and an alteration in muscle fatigue by reducing the duration of contact time (mean value
contraction (Pedersen et al. 2004; Boyas and Guevel 2011). 120 ms). The latter has been reported (Morin et al. 2007) to
Boyas and Guevel (2011) suggested on the one hand that in be the major determinant of leg-spring behavior in active
order to resist the onset of fatigue during prolonged con- athletes not specialized in running. On the contrary, athletes
traction, new motor units are recruited to compensate for A, C and D showed opposite motor behaviors. They
those activated at the start of the contraction. On the other were mainly oriented to positive positions over the first PC.
hand, muscle afferents provide the nervous system with This means that they had long CT (125128 ms), resulting in
information about the state of muscles and appear to be lower values of leg (15.516.58 kN m-1) and vertical
involved in regulating fatiguing exercise. Further investi- (60.9269.49 kN m-1) stiffness than G.
gation is needed to clarify the role of different events
involved in the onset of fatigue during sprinting. The force component
During running, leg stiffness could be considered as a
main cause of possible movement reorganization under the The second pattern was on the second PC of the PCA and
fatigue condition. The kleg is more likely to substitute for linked positively to force and flight time and negatively to

123
Author's personal copy
Eur J Appl Physiol

vertical displacement of the CoM (negative values). It PC. For instance, C1, C2 and C3 were in a positive
characterized athletes H and F who had negative position on the first PC, corresponding to high CT and DL
scores on the second PC, i.e. they relied on high values of values. C4 was in a negative position, corresponding to
force [mean value: H (3.06 kN), and F (2.89 kN)] and high leg and vertical stiffness values.
flight time (respectively, 184.7 and 172.5 ms) during run- In the second case, the athlete adopts a neutral position
ning. The role of Fmax in sprinting has been investigated by during R4. This was the case of athletes B (third PC),
Weyand et al. (2000) who tested the contribution of the D (first PC) and E (third PC). For instance, D1,
ground reaction force compared with contact time and D2 and D3 were in positive positions on the first PC,
stride frequency during treadmill running at different contrary to the position of D4 which was in a neutral
speeds (311.1 m s-1). The authors indicated that running position. That means that the CT of sprinter D decreased
quickly essentially implies greater ground force produc- under fatigue during R4 (Fig. 2).
tion, not greater step frequency or shorter ground contact
duration, which was the case of athletes characterized by a The motor signature remains unchanged with fatigue
force pattern.
In the second behavior, the characteristics of the pattern
The stride component remain unchanged with sprint repetition. In other words,
the athletes were able to maintain their movement orga-
The third individual pattern observed in this study was on nization regardless of the fatigue state. This was the case of
the third PC of the PCA and linked to stride parameters three athletes (F, G and H). The behavior of H
(stride frequency to stride length) during running. The and F, for instance, remained nearly the same with
negative relationship between these two parameters shows repetition of 100-m sprints with negative positions on the
that an increase in stride frequency involves a decrease in second PC (0.35, -0.96 and 0.06, -0.73 correspondingly),
stride length and vice versa. The stride-oriented athletes i.e. with a high level of Fmax and FT. This means that force
scores (B and E) had positive positions on the third and flight time are not affected by fatigue. Fatigue did not
PC, meaning that they relied on a high level of step fre- alter either the sprinters motor behaviors or their corre-
quency [mean value: B (4.04 Hz) and E (4.31 Hz)] and sponding positions on the PC.
consequently showed short stride amplitude (mean value: Two different strategies were identified. The first con-
respectively, 1.99 and 1.98 m). This could demonstrate the sisted in switching from one key variable to another without
importance of stride frequency compared with stride length changing motor behavior. In the second strategy, athletes
in skilled sprinters. maintained their motor behavior with fatigue. Both these
Individual plotting disclosed not only individual motor strategies appear to be efficient in reducing the effects of
signatures among the three PC explained above but also fatigue in skilled sprinters. This confirms that training status
demonstrated the evolution of each pattern with fatigue. is a determinant of efficiency of the SMM parameters as
Two different behaviors of adaptation to fatigue were previously reported in the literature (Bret et al. 2002; Laffaye
identified by the PCA. et al. 2005). Indeed, our results show a slight decrease in
velocity (-3.55%) compared with the literature (-11.6%,
The motor signature is altered with fatigue Morin et al. (2006)) in novices with a similar protocol.
Additionally, the equations found by multiple regression
The first behavior is a change in the pattern with fatigue analysis allow us to understand the connection between the
induced by sprint repetition. This phenomenon was three components. The first PC is linked negatively to the
observed in five sprinters (A, B, C, D and E, other two, explaining that when the contact-time component
see Figs. 2, 3) who maintained their running profile, but decreases, force and stride increase. Second, the relative
changed the key variable of their pattern when exhausted. contribution of each component in both equations, for R1 and
In other words, the motor behavior could be affected by R4, is different. Only the signs of the three components do
fatigue so that the athlete relies more on different param- not change, showing the same relationship between com-
eters than those used in the pre-fatigue state. This change ponents independently of fatigue. Without fatigue, the
was characterized in two different manners: changing the greatest contributor is the force pattern (56%); then it
key-variable of the pattern or adopting a neutral position switches to stride pattern with fatigue, contributing 50% of
on the axis. total variability. This shows that a high level of vertical force
In the first case, the sprinters are likely to be able to is a major factor of velocity, confirming the recent study of
replace the main parameter of their corresponding pattern Morin et al. (2011), but with fatigue, the sprinter is not able to
with another when fatigued. That was the behavior of produce a high level of force but rather seems able to regulate
athletes A and C, whose positions varied on the first the stride component, i.e. stride length and frequency.

123
Author's personal copy
Eur J Appl Physiol

Conclusion Cairns SP (2006) Lactic acid and exercise performanceculprit or


friend? Sports Med 36(4):279291
Cheetham ME, Boobis LH, Brooks S, Williams C (1986) Human
The first aim of this study was to discover whether muscle metabolism during sprint running. J Appl Physiol
mechanical parameters could be affected by the blood 61:5460
lactate concentration in skilled sprinters. The mechanical Chelly SM, Denis C (2001) Leg power and hopping stiffness:
parameters are not correlated to blood lactate rates. The relationship with sprint running performance. Med Sci Sports
Exerc 33(2):326333
latter are correlated only to performance, i.e. to the Delecluse C, Van Coppenolle H, Willems E, Diels R, Goris M, Van
decrease in velocity. The second purpose was to determine Leemputte M, Vuylsteke M (1995) Analysis of 100 meter sprint
the variation of leg stiffness with sprint repetition and in performance as a multidimensional skill. J Hum Mov Sci
the phases of the 100-m sprint. The results show an 28:87101
Dompnier B, Pansu P, Bressoux P (2007) Social utility, social
increase in leg stiffness between R3 and R4. The mainte- desirability and scholastic judgments: toward a personological
nance of leg stiffness and its increase with fatigue were model of academic evaluation. Eur J Psychol Educ 3:333350
determinant for the maintenance of velocity. Our third goal Farley CT, Gonzalez O (1996) Leg stiffness and stride frequency in
was to determine sprint motor pattern behavior as a func- human running. J Biomech 29:181186
Farley CT, Glasheen J, McMAhon TA (1993) Running springs: speed
tion of fatigue induced by four all-out 100-m sprints. The and animal size. J Exp Biol 185:7186
results revealed three sprinting motor signatures. Indeed, Forestier N, Nougier V (1998) The effects of muscular fatigue on the
the PCA shows three components, which explain 88.2% of coordination of a multijoint movement in human. Neurosci Lett
total variance and include the nine studied variables: the 252:187190
Foucart T (1999) Linear multiple regression on canonical variables.
contact-time component, the force component and the Biometr J 41(5):559572
stride component. Gibson ASC, Noakes TD (2003) Evidence for complex system
Adaptations to fatigue are obviously different from one integration and dynamic neural regulation of skeletal muscle
athlete to another. Our study clearly shows the existence of recruitment during exercise in humans. Br J Sports Med
38:797806
three individual running patterns, namely the contact-time Hautier CA, Wouassi D, Arsac LM, Bitanga E, Thiriet P, Lacour JR
pattern, the force pattern and the stride pattern. Each athlete (1994) Relationships between postcompetition blood lactate
was characterized by a sole pattern that could not be replaced concentration and average running velocity over 100 m and
by another under the fatigue induced by sprint repetition. 200 m races. Eur J Appl Physiol 68:508513
Hirvonen J, Rehunen S, Rusko H, Harkonen M (1987) Breakdown of
However, fatigue could constrain some sprinters to replace high-energy phosphate compounds and lactate accumulation
the key variable with another in the same pattern. This during short supramaximal exercise. Eur J Appl Physiol
substitution is likely to be an efficient way of optimizing leg- 56:253259
spring behavior and minimizing the effects on running per- Hobara H, Inoue K, Gomi K, Sakamoto M, Muraoka T, Iso S,
Kanosue K (2010) Continuous change in spring-mass character-
formance of fatigue induced by sprint repetition. istics during a 400 m sprint. J Sci Med Sport 13(2):256261
Kollias I, Hatzitaki V, Papaiakovou G, Giatsis G (2001) Using
principal components analysis to identify individual differ-
ences in vertical jump performance. Res Q Exerc Sport
72:6367
References Laffaye G, Taiar R, Bardy B (2005) The effect of instruction on leg
stiffness regulation in drop jump. Sci Sport 20:136143
Alexander R (1992) A model of bipedal locomotion on compliant McCartney NL, Spriet LL, Heigenhauser GJ, Kowalchuk JM, Sutton
legs. Philos Trans R Soc Lond 338(B):189198 JR, Jones NL (1986) Muscle power and metabolism in maximal
Arampatzis A, Bruggemann GP, Metzler V (1999) The effect of speed intermittent exercise. J Appl Physiol 60(4):11641169
on leg stiffness and joint kinetics in human running. J Biomech McMahon TA, Cheng GC (1990) The mechanics of running: how
32:13491353 does stiffness couple with speed? J Biomech 23(1):6578
Blickhan R (1989) The spring-mass model for running and hopping. McMahon TA, Valiant G, Frederick EC (1987) Groucho running.
J Biomech 22:12171227 J Appl Physiol 62(6):23262337
Bogdanis GC, Nevill ME, Boobis LH, Lakomy HK, Nevill AM Mero A, Komi PV, Gregor RJ (1992) Biomechanics of running.
(1995) Recovery of power output and muscle metabolites Sports Med 13(6):376392
following 30 s of maximal sprint cycling in man. J Physiol Morin JB, Belli A (2003) Mechanical factors of 100 m sprint
15(482 Pt 2):467480 performances in trained athletes. Sci Sports 18:161163
Boyas S, Guevel A (2011) Neuromuscular fatigue in healthy muscle: Morin JB, Dalleau G, Kyrolainen H, Jeannin T, Belli A (2005) A
underlying factors and adaptation mechanisms. Ann Phys simple method for measuring stiffness during running. J Appl
Rehabil Med 54(2):88108 Biomech 21:167180
Bret C, Rahmani A, Messonnier L, Bourdin M, Bedu E, Lacour JR Morin JB, Jeannin T, Chevallier B, Belli A (2006) Spring-mass model
(2001) Relationship between the post-competition blood lactate characteristics during sprint running: correlation with perfor-
concentration and the 100 m sprint running performance. Sci mance and fatigue induced changes. Int J Sports Med
Motr 42:2428 27:158165
Bret C, Rahmani A, Dufour AB, Messonnier L, Lacour JR (2002) Leg Morin JB, Samozino P, Zameziati K, Belli A (2007) Effects of altered
strength and stiffness as ability factors in 100 m sprint running. stride frequency and contact time on leg-spring behavior in
J Sports Med Phy Fit 42:274281 human running. J Biomech 40:33413348

123
Author's personal copy
Eur J Appl Physiol

Morin JB, Edouard P, Samozino P (2011) Technical ability of force stiffness and overall musculoskeletal stiffness during hopping.
application as a determinant factor of sprint performance. Med Eur J Appl Physiol 103(2):163171
Sci Sports Exerc (Epub ahead of print) Rabita G, Slawinski J, Girard O, Bignet F, Hausswirth C (2011)
Nummela A, Vuorimaa T, Rusko H (1992) Changes in force Spring-mass behavior during exhaustive run at constant velocity
production, blood lactate and EMG activity in the 400 m sprint. in elite triathletes. Med Sci Sports Exerc 43(4):685692
J Sports Sci 10:217228 Rehunen S, Naveri H, Kuoppasalmi K, Harkonnen M (1982) High-
Nummela A, Gundersen JS, Rusko H (1996) Effects of fatigue on energy phosphate compounds during exercise in human slow-
stride characteristics during a short-term maximal run. J Appl twitch and fast-twitch muscle fibres. Scan J Clin Lab Invest
Biomech 12:151160 42:499506
Pedersen TH, Nielson OB, Lamb GD, Stephenson G (2004) Weyand PG, Sternlight DB, Bellizzi MJ, Wright S (2000) Faster top
Intracellular acidosis enhances the excitability of working running speeds are achieved with greater ground forces not more
muscle. Science 305:11441147 rapid leg movements. J Appl Physiol 89:19911999
Rabita G, Couturier A, Lambertz D (2008) Influence of training
background on the relationships between plantarflexor intrinsic

123

You might also like