Participants

Six male (24.1 ± 2.1 yrs. 77.6 ± 4.4 kg 180.5 ± 1.7 cm) and eight female (21.7 ± 1.7 yrs. 62.6 ± 7.1 kg 169.7 ± 3.6 cm) cross-country skiers participated in the study. All participants were members of the Norwegian University of Science and Technology student ski team, well-trained in XCS and familiar with roller-skiing on a treadmill. The study was approved by the Norwegian Social Science Data Service (NSD), ref.nr 512269, and the experiments were conducted in accordance with the Helsinki Declaration. The participants signed a written informed consent and were allowed to withdraw from the project at any time without providing a reason.

Protocol

The tests were performed on a motorized treadmill (Forcelink Technology, Zwolle, The Netherlands) using a safety harness. All participants used the same pair of IDT roller-skis (IDT sports, Lena, Norway) equipped with wheels of rolling resistance category 2. A warm-up period assured that rolling friction coefficient was constant (μ≈0.018) [9] during the experiments. The athletes could use their own poles or choose poles provided by the laboratory at length interval of 5 cm.

For each participant, the experiments were performed within one hour, including warm-up. During a ten-minute warm up, the skiers were acquainted with the treadmill and performed classic roller skiing at different inclines and speeds, assuring all techniques were used. They were informed about the general design, i.e., during the exercise they were to roller ski (classic technique) freely, choosing a technique and execution as they saw fit while some backwards pulling resistive force was added or removed at certain time-intervals. All powers were estimated to be submaximal, although the highest ones might have exceeded some individuals’ anaerobic thresholds. They were further informed that each bout (i.e., one speed-incline combination) was not expected to exceed seven minutes. The four combinations of two speeds (10 km h^-1 and 12 km h^-1) and two inclines (5% and 8%) were studied. These combinations were selected based on our previous studies [9, 15], assuring that they would be submaximal efforts, and the expectation that most athletes would not choose DS technique at the onset of the protocol because DS is usually chosen on steeper inclines.

A specially built pulley system was used to guide a rope between the athlete and a hanging weight. Using a broad belt, the rope was attached at the waist, just above the pelvic region, close to the assumed centre of mass location. The height of the pulley system was adjusted so that the pulling force was applied approximately parallel to the treadmill surface at both inclines. Minimal friction of the pulley wheel was assured by using a road-bicycle wheel for that purpose. The weights consisted of small bags filled with sand, and weighing 500 g, that could be manually added to or removed from the rope with ease. A 500g hanging weight corresponded to a ~13–16 W change in external mechanical power. Approximately every 20 seconds an additional weight was added to the rope until the athlete had adopted the DS technique. In such case, one more 500 g resistance mass was added at the same time interval to check if this DS technique was maintained. Occasionally, on the 5% incline, an athlete did not choose the DS technique even after adding considerable resistance. In such cases, the bout was terminated. Before including this outcome in further analysis, it was, after consultation with the athlete, affirmed that the athlete indeed would not have changed to that technique at higher resistance, and it was affirmed that the final obtained power exceeded that of transition to DS at other conditions (also see statistical analysis below).

Recording equipment

A load cell (N-DTS-FS5, Noraxon USA Inc., Scottsdale, Arizona) connected in series in the rope-connection between pulley and athlete recorded the actual resistive force at 1500 Hz. Kinematic data were collected with Oqus 3D motion capture (Qualisys AB, Gothenburg, Sweden) using eight cameras at 200 Hz. Markers at the tip end of the poles and on the skis were used to identify the techniques using a custom algorithm [15] in Matlab (R2019b, Mathworks Inc., Natick, MA, USA).

The experiments were performed at the core facility NeXt Move, Norwegian University of Science and Technology (NTNU).

Signal treatment and statistical analysis

Both resistive force and power were statistically analysed as potential triggers. While resistive force was recorded directly, the external mechanical power was estimated according to [20] and adding the power required to overcome the resistive force (F_p): (1) Where m is effective body mass (including ski equipment), v is treadmill belt speed (‘velocity’), g gravitational acceleration, α angle of incline, μ coefficient of rolling friction, F_p the force on the pulley rope. Note that the different incline-speed combinations result in different resistive force-power combinations, allowing us to distinguish the role of power from the mechanism that it was induced with (additional external resistive force). Even though our main aim was to investigate the role of power, it was also possible to elucidate the means by which power was altered, i.e., the resistive force. All athletes underwent the same resistive force increment despite considerable differences in body mass. Thus, there were considerable interindividual differences in total resistance, i.e., including gravity. Therefore, alongside additional resistive force, total resistive force (i.e., pulley force, gravitational component, and roller friction: mg(sinα + μ cosα)+ F_p was also investigated as a possible trigger. It should be noted that, in this study, only the mean external mechanical power is examined; any fluctuations in power due to the motion of limbs relative to the centre of mass, mechanical energy fluctuations of the body and associated losses were not considered. Because all classic XCS techniques include at least one propulsion and one glide phase during each stride, the additional resistive force fluctuated considerably during one stride cycle. Therefore, force and power time traces were smoothed using a moving average with a time window of 2 seconds. Force and power at the time of technique transition was averaged over a period of –1.5 to +1.5 seconds (i.e., period of about 1–2 whole stride cycles) around that time point. The main outcome of this procedure is shown in Fig 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Equipment setup, study design, and original time traces (cycle rate, power and additional resistive force) during one session.

A. Athlete on treadmill with hanging resistance weight. B. Timeline of protocol showing resistance weights. C. Cycle rate and power time trace (at v = 10 k h^-1, 8% incline). The techniques are indicated by colour. Bottom diagram shows the raw resistive force data (also in zoomed range) and smoothed over a 2 s window.

https://doi.org/10.1371/journal.pone.0283794.g001

The main statistical tests were comparisons between the external resistive forces and mechanical powers at the technique transitions across the two incline and two speed conditions. If force and/or power is to be regarded as an isolated trigger factor, the force and/or power at which transitions occur should stay constant across different incline-speed combinations. Thus, any difference in the external resistive force or mechanical power at the technique transitions between the four incline-speed conditions was regarded as evidence that the variable is not an isolated trigger factor. In some cases, an athlete did not choose all three techniques (i.e., they did not choose DP or DS) in all four conditions. In such a case, obviously a DP-DK or DK-DS transition was absent. Thus, even though the athlete exhibits behaviour (i.e., not changing technique) that differs from the behaviour in another condition, i.e., necessary and valuable information is available, it would initially appear as ‘missing value’ (no number was associated with such behaviour). To allow us to use such ‘missing value’ in the statistical analysis, it was filled in with a fictitious value, one force or power step higher than the maximal value or one step less than the minimal value observed for that athlete during all four incline-speed settings. A Bayesian 2x2 ANOVA for repeated measures (RM) was used to estimate means (± 95% credible intervals, CI). We report Bayes factors for inclusion (BF_in), quantifying the evidence (odds) for the model including incline or speed as predictors given the data (i.e., change from prior odds (0.5) to posterior odds [see e.g., 21–23]). We also report classical F- and p-values (2x2 RM ANOVA) as well as partial eta squared (n²). The Bayesian approach quantifies evidence both in support of (the power and force at transition are equal in different conditions) and against the null-hypothesis (the transition power and—force differ between conditions). To avoid dichotomization of evidence and both type I and type II errors, we used a more ‘holistic’ approach when judging whether force and power at technique transition change or not [see e.g., 24]. Therefore, we focus both on BF_in, p-values but also on the actual differences and spread (95%CI), as well as data quality. For the Bayesian ANOVA, the prior for fixed effects (probability of incline or speed as predictors) was set at 0.5 [25]. For the main tests (effect of incline and speed on power and force at transition), the assumption of approximately normally distributed residuals was confirmed by checking Q-Q-plots visually.

To test for technique transition hysteresis, the external resistive forces at the technique transitions were compared between when the force was being incremented vs. decremented using Bayesian paired t-tests. BF₁₀, i.e., the relative likelihood of the alternative versus the null hypothesis, is reported with a Cauchy (r scale 0.707) prior [26]. RM ANOVA and paired t-tests were used in the descriptive analysis of the dataset. All statistical analysis were done using JASP (JASP Team 2022. JASP (Version 0.16.1) [Computer software]).

Potential triggers for technique transition

Regarding potential external triggers for technique transition, it should be noted that we tested the null hypotheses that technique transitions would occur at similar resistive forces and power outputs regardless of incline or speed. The challenge of this approach was explained earlier [9] and the interpretation of the classical statistical outcome (p-values) is based on the process of elimination: any factor (resistive force or power output at transitions) showing significant differences between conditions is not a trigger, but the other factors might be. In this regard, Bayes factors are helpful as they provide evidence, both for and against, the null hypothesis, i.e., equality of power and force at transition, independent of condition. Keeping this in mind, our interpretation is as follows. 1: In conditions where only speed was changed, resistive force (both additional and total) may be a trigger, but power is not. Comparing transitions at the same incline but different speeds, athletes tend to shift technique at same resistance, not the same power (Fig 4, Table 1). Referring to the Bayes factor values for the speed effect on force at transition, it should be noted that only anecdotal evidence is available supporting the null-hypothesis (identical force at transition) compared to the alternative (Table 1: BF_in = 0.5–0.9 –a value of at most ~0.2 would suggest otherwise). In other words, we cannot conclude that resistive force definitively has a trigger function, but the possibility for such is present. 2: In conditions where only incline was changed, neither force nor power is a clear candidate for this trigger function (possibly with the exception of total resistive force in DP-DK transition) (Fig 4, Table 1). These findings seem to agree with earlier studies: Both Ettema, Kveli [15] and Løkkeborg and Ettema [9] indicated that incline seemed to play a bigger role for technique transitions than did speed or power. Yet, as mentioned, the design of these studies did not allow any conclusion concerning power as such a factor in isolation, i.e., obtained independently of incline and speed.

The role of external power

The current results (Fig 4, Table 1) confirm that external mechanical power generally cannot be regarded as a trigger for technique transitions. External power is calculated from a combination of speed, incline and resistive force. In contrast to our previous studies [9, 15], here power was manipulated independent of speed and incline. Still, the most likely conclusion supports our previous findings: rather than mechanical power output itself, one or more of the underlying factors (e.g., external resistance) may be a trigger for technique transitions. The marginal role of power in the choice of technique is substantiated by the fact that in this and previous studies, all techniques were chosen even though intensity was only moderate. Further, in XCS competitions, when intensity is clearly very high, all three techniques are chosen under different conditions (terrain, snow friction, wind etc.) [19, 27].

The finding that external power itself unlikely functions as a trigger for technique transitions is comparable to those found in studies on the walk-run transitions. Even though people tend to use the gait (walking vs. running) that is energetically most economical, energetics are not a direct trigger to transition between gaits. The change in movement economy (i.e., V̇O₂) changes too slowly to be considered a trigger for an abrupt gait transition [e.g., 2, 4, 6, 28]. As in walk-run transitions, other variables that do change more abruptly (i.e., muscular processes—the origin of V̇O₂ changes) are more likely triggers, though not easily measured.

Potential mechanisms

The role of resistance force, used as mechanism to change power independent of speed or incline, was investigated as secondary aim of study because it potentially could assist elucidating mechanisms that may be responsible for choice of technique. In contrast to power, almost identical resistive forces were found on average at DK-DP and DK-DS transitions between speeds at the same incline, but there were considerable interindividual differences. The power jumps per weight added differed by only 3 W at the two treadmill speeds. Thus, this clear discrepancy between power and force concerning the effect on behaviour can hardly be explained by this small protocol anomaly. Therefore, external resistive force may play a trigger role on a fixed incline. As mentioned, there is no evidence for either outcome (‘force definitively has no trigger role’ vs. ‘force clearly has a trigger role’). Yet, the difference between power and additional resistive force in this respect is distinct. These findings strengthen the notion of the strong trigger function of incline—and possibly the associated mechanism of using the body’s mechanical energy in ‘falling’ on the poles that is more effective at shallower inclines [see 9, 18]. Beyond that, a direct trigger role of resistive force (when incline is not in play) does not necessarily indicate that the sense of resistance (in this case around the pelvic region) is the intrinsic trigger. When maintaining speed, as was done in all experiments, total resistive force is uniquely related to propulsive force. The sense of having to apply increasing propulsive force is the corresponding internal trigger candidate.

Moreover, we cannot induce any general principle about resistive force, i.e., independent of the way additional resistive force was applied (here close to centre of mass). In this respect, the current study cannot elucidate if additional or total resistive force is the key for such ‘sense of effort’. Still, the fact that additional force and total force (and power) at transition show opposite effects of incline (Fig 4) suggests that the way resistance is applied is plays a role in this ‘sense of effort’ leading to the choice of technique. However, as mentioned, the current study was not designed to investigate this detail explicitly.

The current findings support the notion presented by Pellegrini, Zoppirolli [16], i.e., maintaining (pole) forces—which are directly linked to resistance forces in DP as applied in this study—below a certain threshold serves as a trigger. In the present study, athletes changed from DP, via DK, to DS with increasing demand for propulsion, independent of incline. In previous studies [9, 15], athletes did the same, but the increased (propulsive force) demand was induced by increasing incline. In DP, propulsion through the skis is not possible. The only way for power from the legs to contribute to propulsion through the poles is by the utilization of body’s mechanical energy, i.e., “falling on the poles”. Therefore, in those studies, we opted for the notion that this mechanism becomes less effective at steeper inclines [17]. This leads to the choice of DS because the legs can generate considerably more power directly through ski push-offs [17].

The current findings suggest that at least one other mechanism contributes to triggering the transition to DS on steep inclines. We cannot decipher from our data if indeed a maximal threshold for poling force plays a key role [16]. However, the notion that poling force acts as a trigger seems plausible. Even though the lower extremities can contribute profoundly to power production in DP [18, 29–31], all propulsive force ultimately is applied via the poles, placing considerable demands on the upper extremities. At the higher resistances induced by incline, upper extremity muscle activity (and the time of force generation) may become too great [18] and by transitioning to DS, the propulsive force can be distributed between both poles (~arms) and skis (~legs). Thus, while we previously found evidence against a pole force trigger [17], the current study—using a different protocol paradigm—to some extend supports the pole force trigger hypothesis. This idea is supported by Andersson, Pellegrini [32] who found that pole forces in DS increased with increasing speed, likely related to a reduced propulsion time [32] (and may explain why DS usually is not chosen at high speeds).

Hysteresis

Based on our previous studies [9, 15], we expected hysteresis to occur because power changed during the protocols. This expectation was confirmed, and once more the hysteresis effect was rather small but relatively consistent. Our present data suggest hysteresis not only occurs when incline is changed but applies to a wider ‘terrain’ of conditions. Like in walk-run gait transitions, hysteresis is an inherent and complicating aspect of technique transitions in XCS.

Atypical behaviour

Curiously, one of the athletes never used—and another one only used—DS. No obvious difference in demographics or experience was noted for these athletes. Obviously, they were not included in the statistical analysis for transitions they never made. This also applies for a few other athletes who used only one technique in some of the conditions. This is the reason for the discrepancy between number of athletes that participated and the n-value for the analysis (14 vs 11 and 10), which was not due to technical errors. The behaviour of these athletes should not be disregarded in the interpretation of the findings. It rather highlights the complexity of the mechanism behind technique transitions. Likely, for these athletes, the range of the speed-incline-resistance conditions was outside of their individual range for one or two of the techniques.

Methodological considerations

Like in our previous studies, we made stepwise changes in one of the external conditions, in this case external resistive force. It is not clear if a continuous change would have affected the outcome of study. However, as indicated previously [9], we have no evidence of that being the case. The identification of the exact resistive force and power at spontaneous transitions would have been more precise, but not essentially different. The outcome of this study confirms that the current sensitivity was sufficient for detecting the role of power. A study with higher force change resolution, i.e., smaller weight increments, may elucidate if indeed force is (close) to identical for DK-DP transitions at different speeds, or whether small force differences—undetectable in the present study—occur.

The choice of speeds and inclines was based on previous studies [9, 15], but resulted in larger power differences between the two inclines than the two speeds (at the same resistance force). This raises the question if speed differences would have been somewhat larger, would the statistical outcome for speed have resembled more closely that for incline (Table 1) and disqualified resistance force as trigger for techniques transition as well? One may argue that this would not have an influence since power itself is not a factor of importance regarding triggering technique transition. In any case, the current findings are not conclusive about the role of resistance force.

The continuous recording of resistive force (Fig 1) shows large force fluctuations (and thus power fluctuations). On might argue that these large fluctuations have affected the outcome of this study. However, the way resistance was applied may only have amplified such fluctuations, not caused them. Power fluctuations are inherent to the XC skiing techniques. For example, in this study, the body’s speed changes during one cycle amounted around 1 m s^-1, which is associated with about 100 W power fluctuations (including potential and kinetic). The total power fluctuations registered were up to about 200 W, caused by fluctuations in additional resistive force (from the pulley). In other words, regardless the locomotion form, power fluctuation during one propulsion cycle is an inherent aspect of locomotion, and definitively in classic cross-country skiing [17].

Our sample size was relatively small (N = 14). Thus, some robustness assessment was warranted. Therefore, we used JASPs robustness check software to see how statistical outcomes depended on prior width and N. Although statistical outcomes are vulnerable at low N, for no variable was our conclusion affected in any way when reducing N down to 7 or 8.

Concluding remarks

In all, considering the previous studies and the current one, the notion that a single factor may act as trigger for transition seems naïve. However, based on the present study, we can now reject external mechanical power demand as an important trigger for cross-country skiing technique transitions. The ‘sensation’ of a technique becoming unsuitable for the external conditions at hand, likely has a multifactorial origin, including local muscle forces, (visual [33]) perception of speed and the gravitational field [see also 16]. In a similar way, walk-run transition speed is affected by incline [2], indicating that speed itself (or any directly associated factors) cannot be the sole external factor triggering gait transitions.