The training intensity distribution among well-trained and elite endurance athletes

Researchers have retrospectively analyzed the training intensity distribution (TID) of nationally and internationally competitive athletes in different endurance disciplines to determine the optimal volume and intensity for maximal adaptation. The majority of studies present a “pyramidal” TID with a high proportion of high volume, low intensity training (HVLIT). Some world-class athletes appear to adopt a so-called “polarized” TID (i.e., significant % of HVLIT and high-intensity training) during certain phases of the season. However, emerging prospective randomized controlled studies have demonstrated superior responses of variables related to endurance when applying a polarized TID in well-trained and recreational individuals when compared with a TID that emphasizes HVLIT or threshold training. The aims of the present review are to: (1) summarize the main responses of retrospective and prospective studies exploring TID; (2) provide a systematic overview on TIDs during preparation, pre-competition, and competition phases in different endurance disciplines and performance levels; (3) address whether one TID has demonstrated greater efficacy than another; and (4) highlight research gaps in an effort to direct future scientific studies.


INTRODUCTION
The intensity and duration of work as well as recovery periods define overload and adaptations in athletes (Faulkner, 1968). While there appears to be consensus regarding the factors that limit endurance performance (Joyner, 1991;Coyle, 1995;Hawley and Stepto, 2001), agreement regarding the optimal volume and training-intensity distribution (TID) among elite athletes remains elusive. Achieving such consensus is important in order to maximize training adaptations and translate them into performance gains while avoiding overtraining.
Researchers have generally employed retrospective designs to analyze the TID of nationally or internationally competitive athletes in different endurance disciplines. In contrast, the number of prospective quasi-experimental or experimental studies investigating athlete responses to different TID's are small, with only limited studies examining well-trained or elite endurance athletes (Evertsen et al., 1997(Evertsen et al., , 1999(Evertsen et al., , 2001Billat et al., 1999;Ingham et al., 2008Ingham et al., , 2012Yu et al., 2012;Stöggl and Sperlich, 2014). Articles reviewing the training intensity and duration of endurance athletes (Seiler and Tonnessen, 2009;Seiler, 2010) conclude that: (1) elite endurance athletes perform approximately 80% of their training at low intensity (<2 mM blood lactate) with about 20% structure (Holmberg, 1996), in which the majority of training time is spent in HVLIT ("zone 1"), and a decreasing proportion of training time in zones 2 and 3.
Since nearly all studies dealing with TID were based on retrospective analysis, we recently employed a randomized controlled design to investigate which TID (HVLIT vs. THR vs. HIT vs. polarized) provided the greatest response on key components of endurance performance among well-trained KEY CONCEPT 7 | Key components of endurance performance In connection with many endurance sports five key parameters are utilized for comparison of performance: (1) peak oxygen uptake; (2) velocity or power output at the lactate threshold; (3) work economy; (4) peak running velocity or peak power output, and (5) time to exhaustion.
athletes (Stöggl and Sperlich, 2014). We concluded that the polarized TID resulted in the greatest improvements in the majority of key endurance performance variables assessed, and THR or HVLIT did not lead to further improvements in performance. However, as numerous retrospective reports have shown conflicting results, the question regarding which TID represents the "best-practice" model for inducing performance gains-while avoiding overtraining-remains open to debate. Therefore, the aims of the present review were to: (1) summarize the main responses of different retrospective and prospective studies exploring TID; (2) provide a systematic overview of TIDs during preparation, pre-competition, and competition phases in different endurance disciplines and performance levels; (3) address whether one TID has demonstrated enhanced efficacy over another; and (4) highlight research gaps in an effort to direct future scientific studies.

Retrospective Studies
One major problem in TID investigations lies in the difficulty of involving elite athletes in a scientific experiment. Given their already high fitness levels, introducing certain novel training programs among elite performers may not result in performance enhancement and can even lead to overtraining symptoms. Therefore, the majority of studies dealing with TID in welltrained to elite endurance athletes are based on retrospective analyses of their training (Tables 1, 2, Figure 1).

Exercise-intensity Distribution during the Preparation Period
From 1990 to 2014, the TID of elite, nationally ranked to worldclass athletes who were training in their preparation phase were reported. These athletes competed in rowing (Hartmann et al., 1990;Guellich et al., 2009), running (Robinson et al., 1991), cycling (Lucia et al., 2000), and cross-country skiing (Sandbakk et al., 2011;Tonnessen et al., 2014). Findings indicate that elite endurance athletes spend a high percentage of their TID in a  FIGURE 1 | The training-intensity distribution (i.e., percentage time spent in zone 1: <first ventilatory threshold or steady-state lactate at ∼2 mM; zone 2: at or near lactate threshold (∼4 mM) or second ventilatory threshold; zone 3: high-intensity training above lactate or second ventilatory threshold) in well-trained to elite endurance athletes in retrospective analyses during (A) preparation phase, (B) pre-competition phase, (C) competition phase, and (D) seasonal analysis.
In professional cyclists (Lucia et al., 2000) and top-class runners (Billat et al., 2001) engaged in pre-competition training, similar proportions of HVLIT were reported (78%). The distribution of zones 2 and 3 were however, polarized (4 and 18%) in the runners and pyramidal (17 and 5%) in the cyclists.

Exercise-intensity Distribution during the Competition Phase
Documentation of the TID during the competition period is rare since (a) technical equipment may not be applied during competition, (b) the TID largely depends on the amount and type of competitions (e.g., single races vs. stage races), and (c) the strategies for tapering for competitions vary widely across sports. Lucia et al. (1999) reported a pyramidal TID (70/23/7%) during the Tour de France based on the "HR time in zone" method over 22 competition days. The exercise intensity was particularly high during the time trials and high mountain stages. Also Lucia et al. (2000) reported that elite cyclists performed approximately 810 km·wk −1 (May) with a TID of 77/15/8%, while elite crosscountry skiers and biathletes (Tonnessen et al., 2014) showed a higher proportion of HVLIT compared with THR and HIT (∼87.5% zone 1-2 vs. ∼12.5% zone 3-5) when compared with the cyclists (Figure 1C).
Comparable with the TID in the cycling studies during the pre-competition phase, regional-and national-class Spanish runners (4-5 h·wk −1 ) demonstrated a pyramidal TID of 71 (<VT 1 ), 21 (VT 1 -VT 2 ), and 8% (>VT 2 ) (Esteve-Lanao et al., 2005) over a 6 month period. The TID of national and international-level swimmers revealed a pyramidal TID (although the athletes spent almost the same time in zone 2 and 3) over an entire season (77/12/11%) (Mujika et al., 1995). The Norwegian elite cross-country skiers and biathletes analyzed during the year leading to their most successful career competition   (Tonnessen et al., 2014) spent 91% of their training time in zones 1-2 and 9% in zones 3-5 or 77 vs. 23% when applying the session goal approach. The monthly frequency of HIT sessions and "zone 5" sessions increased from the general to the specific preparation period and remained unchanged within the competition period. From the end of the general preparation to the peaking phase, the amount of HVLIT decreased by 21%, and HIT-especially zone 5-increased by 40%. Therefore, the TID changed from an emphasis on HVLIT during preparation, toward a pyramidal TID during precompetition, and a polarized TID during the competition phase.
Exercise-intensity Distribution during Long-term Analysis (>1 Year) (Table 2) The TID across several decades was reported in rowers (Fiskerstrand and Seiler, 2004) and speed skaters (Orie et al., 2014). During the 1970s, the training volume of elite rowers was almost equally divided between HVLIT and HIT sessions (Fiskerstrand and Seiler, 2004). Then two major changes across time were identified: (1) training with low blood lactate (<2 mM) increased from 30 to 50 h·month −1 and race pace and supramaximal intensity training (8-14 mM) decreased from 23 to 7 h·month −1 , and (2) total training volume increased from 924 (600-1020) h·yr −1 during the 1970s to 966 (840-1140) h·yr −1 in the 1980s, and to 1128 (1104-1200) h·yr −1 in the 1990s (∼20% increase). Further increase in total training volume in the 1990s did not lead to further improvement in physical capacity.

Prospective Studies Investigating TID
Single Case or Quasi-experimental Longitudinal Studies without Control Groups (Table 3) Based on three studies, we conclude that an increase in HVLIT at the expense of THR leads to performance enhancements, while the exaggerated increase in HIT at the expense of HVLIT might be applied with caution. Billat et al. (1999) examined whether one HIT session compared to three HIT sessions·wk −1 is sufficient to improve performance in four middle-distance runners. The implementation of four HVLIT, one HIT, and one THR session over 4 weeks resulted in improved running speed at maximal oxygen uptake (VO 2max ) and running economy. A further 4 weeks intensification, including two HVLIT, three HIT, and one THR session each week, showed no additional performance benefit, but increased subjective muscle stress, reduced sleep quality, and increased plasma epinephrine, all indicators of impending overtraining.
Altering TID from a THR-emphasis toward a more polarized or pyramidal TID showed improvements in competition performance and physiological capability. In the case study of a male international 1500 m runner (PB 3:38.9 min:s), HVLIT was performed within the first year above the coach's prescribed level and "tempo" training at an excessively high intensity. In the second year, HVLIT (<80% VO 2max ) increased from 20 to 55% and intense training at 80-90% VO 2max and 90-100% VO 2max was reduced from approximately 42 to 20% and 20 to 10%, respectively. Furthermore, a concomitant increase in the proportion of HIT (100-130% vVO 2max ) from 7 to 10% was observed. This training modification coincided with improvements in physiological capability (increase in VO 2max from 72 to 79 ml·min −1 ·kg −1 ) and performance improvements (3:38.9 to 3:32.4 over 1500 m) (Ingham et al., 2012).
The 1 year adaptation of a THR emphasized (41/51/7%) vs. a polarized (86/5/10%) TID were evaluated in nine Chinese top-level sprint speed skaters (500 and 1000 m) and their performances at national competitions, World Cups, and the Olympics (Yu et al., 2012). The overall training durations and frequencies were similar across the two seasons, with THR constituting 50% of the training time (41/51/7%) and with POL 1 year later only 5% (86/5/10%). The increase in HVLIT, the reduction of THR from 50 to 5%, and the increase in HIT led to 2-4% improvements in the 500 m and 1000 m events. (Table 4) The focus of the nine studies manipulating TID was mainly to compare the change in performance and/or adaptation to three different TID's including: THR-emphasized training, HVLITemphasized training, and polarized TID. In the majority of studies, recreational or sub-elite athletes were investigated. All experimental studies reported increased endurance performance, however, in most of the studies the polarized or HVLIT-emphasized TID resulted in the greatest responses of various endurance performance variables (Esteve-Lanao et al., 2007;Ingham et al., 2008;Neal et al., 2013;Munoz et al., 2014;Stöggl and Sperlich, 2014).

Randomized Controlled Studies
Esteve-Lanao et al. (2007) were among the first to conduct experimental randomized controlled studies assessing the effects of 5 months increased or decreased HVLIT on endurance performance. Twelve sub-elite Spanish runners were randomly assigned to two separate groups performing equal amounts of HIT (8.4% of training > VT 2 ). Although the two groups varied in the amount of HVLIT (group 1: 81%, group 2: 67%) vs. THR (group 1: 12%; group 2: 25%), they achieved equal total training loads (TRIMP scores). A polarized TID (74/11/15%) in the group with emphasized HVLIT was revealed with the session goal approach. The improvement in 10.4 km running was greater in the group emphasizing HVLIT (-157 vs. -122 s). If there is sufficient training time, it is advised to design an "easy-hard" rather than a "moderately high-hard" load distribution training approach.
zone 2 in the THR group, the polarized TID showed greater improvements (+7.0%) compared with THR (+1.6%). Stöggl and Sperlich (2014) explored the response of HVLIT (83/16/1%) vs. THR (46/54/0%) vs. HIT (43/0/57%) vs. polarized TID (68/6/26%) on key components of endurance performance in 48 well-trained runners, cyclists, triathletes and cross-country skiers. While all four groups increased time to exhaustion, the polarized TID increased VO 2peak (+11.7%), time to exhaustion (+17.4%), and peak performance (+5.1%) to the greatest extent. Performance at 4 mM increased after polarized TID (+8.1%) and HIT (+5.6%), with no change in the other groups. HIT resulted in decreased body mass (-3 kg) and increased VO 2peak (+4.8%). Exclusive emphasis of THR or HVLIT did not lead to further improvements in endurance performance in well-trained athletes. Evertsen et al. (1997Evertsen et al. ( , 1999Evertsen et al. ( , 2001 randomly assigned 20 welltrained Norwegian cross-country skiers for 5 months to HVLIT vs. a HIT/THR emphasized TID. The HVLIT group spent 86% at an intensity < 1.5 mM (7 sessions·wk −1 with an increase from 10 to 16 h·wk −1 ) and 2-3 sessions·wk −1 HIT (14%), while the HIT/THR group demonstrated 83% HIT/THR at 3-4 mM (12 h·wk −1 ) and 17% HVLIT as recovery. Despite a 60% increase in training volume in the HVLIT group and approximately four times more training at intensity close to LT in the HIT/THR group, physiological and performance changes were modest in both groups. In contrast to other studies (Ingham et al., 2008;Stöggl and Sperlich, 2014), greater improvements in performance (e.g., running speed at LT and performance in a 20 min run) were found in the HIT/THR group compared with HVLIT. Seiler et al. (2013) analyzed the performance adaptations of different types of high intensity interval training. Thirtyfive recreational cyclists were randomized to four training groups with equivalent training the two previous months (∼6 h·wk −1 , ∼1.5 session·wk −1 ). The HVLIT group trained 4-6 sessions·wk −1 , and the three HIT groups trained two sessions·wk −1 with either 4 × 4 min (94% HR max ), 4 × 8 min (90% HR max ), or 4×16 min (88% HR max ) plus 2-3 sessions·wk −1 HVLIT. The 4 × 8 min interval group induced greater average gains in VO 2peak , power at VO 2peak , and power at 4 mM. Subjects without interval training experience before the intervention tended to achieve greater average improvements in VO 2peak , peak power output, and power at 4 mM compared with subjects reporting 1-1.5 HIT sessions·wk −1 . All three interval training groups tended to improve in physiological capacity after the training period, while the HVLIT group remained relatively unchanged (with the exception of a significant increase in power at 4 mM), despite similar or slightly higher total training volumes (8.5 h·wk −1 vs. 5.7-7.6 h·wk −1 ).

METHODOLOGICAL CONSIDERATIONS
Although the authors are aware that the investigations summarized in this review vary regarding the endurance disciplines, athletic level, duration of observation, time of season (preparation, pre-competition, or competition phase) and TID methodology, the present data show no uniform TID pattern among the different endurance disciplines (Figure 1). The methodology incorporated in the retrospective analysis to compare the TID between disciplines might produce discrepancies in numbers. The intensity of endurance exercise is frequently defined as percentage of HR max or VO 2max or blood lactate concentration. The percentages of for example, VO 2max have been shown to lead to inhomogeneous metabolic strain as indicated by the large variability of blood lactate responses (Scharhag-Rosenberger et al., 2010), and therefore may lead to imprecise assignment to the intensity zones during prolonged exercise.
Elite endurance athletes also implement strength training and speed training in their conditioning training. Since these training forms incorporate short (<30 s), very intense, anaerobic bouts, the continuous measurement of for example HR (due to inertia of the cardio-respiratory system at the onset of intense exercise) may preclude exact quantification of the intensity zone for this part of training. In this context, other methods, such as the "session RPE method" or the "session goal approach" (Seiler and Kjerland, 2006) might be applied. For other candidate biomarkers to quantify training load and understand fatigue in athletes we refer to Halson (2014).

INTENSITY DISTRIBUTION AMONG ENURANCE ATHLETES
Most retrospective studies report a pyramidal TID, with extensive HVLIT (>70%), less time in zone 2, and very little time spent in zone 3, independent of the time of season. Three studies on polarized TID involving cross-country skiers during different phases of the season (Seiler and Kjerland, 2006;Sandbakk et al., 2011;Tonnessen et al., 2014) and one on marathon runners (Billat et al., 2001) were found.
Athletes favor HVLIT, since when the training volume is high, low intensity training (<2 mM or ∼55-85% HR max ) is more tolerable (Hartmann et al., 1990). Despite athletes' preference for low intensity training the majority of coaches favored higher intensity [2.5-4 mM (THR)], mainly for theoretical reasons (e.g., Fritsch, 1985(e.g., Fritsch, , 1986Nolte, 1986). Since the amount of HVLIT has been linked to improved race performance (Hagerman and Staron, 1982;Steinacker, 1993;Esteve-Lanao et al., 2005;Seiler and Kjerland, 2006), the necessity of HVLIT in achieving physiological adaptations for gains in performance has been pointed out in longitudinal observations and experimental designs (Esteve-Lanao et al., 2005Zapico et al., 2007;Ingham et al., 2008). However, when the amount of HVLIT by elite athletes is doubled, no further improvement in performance is evident (Costill et al., 1991), and the athletes mood may be negatively affected (Raglin, 1993). Therefore, for elite endurance athletes with high amounts of HVLIT, the ability to distribute the training intensity optimally may be paramount to both success and counteracting non-functional overreaching (Fiskerstrand and Seiler, 2004).
Runners who have prioritized HIT instead of THR into their HVLIT-training have been reported to perform better (Billat et al., 2003), a result which was corroborated in a prospective study (Stöggl and Sperlich, 2014). Approximately two HIT sessions·wk −1 have been proposed to stimulate performance adaptations without inducing chronic stress (Seiler, 2010). While it was shown that an increase from one to three HIT sessions per week was not accompanied with further performance benefits, such an increase did result in greater subjective muscle stress, plasma epinephrine, and reduced sleep quality, all indicators of impending overtraining (Billat et al., 1999).
Although there are numerous time-efficient adaptations and health benefits associated with HIT, there is evidence that individuals will need to feel physically capable and adequately motivated to perform and maintain high intensity exercise (Hardcastle et al., 2014). Additionally, condensed HIT over a longer period (9 weeks) may lead to a loss in body mass in well-trained athletes (Stöggl and Sperlich, 2014) which may also negatively impact health.
Training adaptation is subject to high inter-individual response , and disciplines with high eccentric forces, high force impacts (e.g., running), and chronic muscle damage will not necessarily exhibit similar TID when compared to disciplines with lower eccentric impact (e.g., swimming, cycling, rowing) because recovery and low-intensity exercise might be less prominent. From this point of view the same TID will in all likelihood induce different adaptations among individuals, even if they behave and exercise in an "elite" manner.
Although the number of retrospective studies reporting a HVLIT or pyramidal based TID is substantial compared to polarized TID in well-trained to elite endurance athletes (Tables 1-4, Figure 1), the findings from various prospective studies (≤5 months training intervention), suggest that a polarized TID results in superior training and performance responses compared to HVLIT and THR in some endurance disciplines and certain phases of the season. Since, variation of the training stimuli is a critical aspect of effective training (Kiely, 2010(Kiely, , 2012, switching from a long-term unidirectional training stressor (e.g., HVLIT) to another training stressor that provides a substantial increase in the amount of HIT (e.g., polarized TID) may prove fruitful. The optimal type of periodization model however, remains open for debate (Issurin, 2010;Kiely, 2010Kiely, , 2012. For instance, it is noteworthy that the effects of an inverse pyramidal or inverse polarized TID (applying a high amount of HIT with lower portion of HVLIT-e.g., 20:80), or exclusive HIT for a longer period (>9 weeks) has not been analyzed. Based on the experience of our previous investigation (Stöggl and Sperlich, 2014), researchers-especially those conducting prospective studies-will have to face the question which TID (in combination with different periodization models) is superior in athletic and health seeking populations.
Technical advancements integrating different internal (e.g., HR, oxygenation via near-infrared spectroscopy, sleep analysis, breathing pattern, HR variability, hand-held analysis of saliva and blood, questionnaires, etc.) and external (power output, GPS-based distance and speed, accelero-and inclinometer, etc.) sensor technologies that enable biological monitoring at a distance (Chan et al., 2012) will further enhance the estimation of individual athletes' optimal TID in a timely manner. Yet, the question remains: which data are the best foundation for quantifying TID?
In summary, most retrospective studies on well-trained to elite endurance athletes report a pyramidal TID, with a large proportion of HVLIT. Polarized TID has been proven to be an effective strategy for some elite athletes during certain phases of the season. However, experimental studies lasting 6 weeks to 5 months demonstrate superior responses to polarized TID, especially when compared with TID that emphasizes THR or HVLIT. As pointed out, the combination of HVLIT with HIT may improve endurance performance with potentially less autonomic and hormonal stress and boredom. The reasons for the non-uniform TID among endurance disciplines may arise from differences in methodology in retrospective analyses and/or high inter-individual variation in the training response. Furthermore, the long-term effects of different forms of TID (e.g., inverse pyramidal or inverse polarized or exclusive HIT) with different patterns of periodization on well-trained to elite endurance athletes, have yet to be characterized. Consequently, an "optimal" TID cannot be identified, and future prospective randomized investigations conducted over extended timeperiods will have to be designed to address this question.