Sex Differences in Ultra-Triathlon Performance at Increasing Race Distance

It has been argued that women should be able to outrun men in ultra-endurance distances. The present study investigated the sex difference in overall race times and split times between elite female and male Ironman Triathletes competing in Ironman Hawaii (3.8 km swimming, 180 km cycling, and 42.195 km running) and Double Iron ultra-triathletes (7.6 km swimming, 360 km cycling, and 84.4 km running). Data from 20,638 athletes, including 5,163 women and 15,475 men competing in Ironman Hawaii and from 143 women and 1,252 men competing in Double Iron ultra-triathlon races held worldwide between 1999 and 2011 were analyzed. In Ironman Hawaii, the sex difference in performance of the top three athletes remained unchanged during the period studied for overall race time. For Double Iron ultra-triathletes, the sex difference for the top three athletes remained unchanged for overall race time. Sex differences increased as endurance race distances increased and showed no changes over time. It appears that women are unlikely to close the gap in ultra-endurance performance with men in ultra-triathlons in the near future. Physiological (e.g., maximum oxygen uptake) and anthropometric characteristics (e.g., skeletal muscle mass) may set biological limits for women.

mances of women and men in diff erent sports. In every discipline, male performance was better than female performance with a mean sex diff erence of 10.0% (SD = 2.9).
While some studies have suggested that women might be able to reach men's endurance performance levels (Whipp, & Ward, 1992;Speechly, Taylor, & Rogers, 1996;Bam, Noakes, Juritz, & Dennis, 1997;Hoff man, 2008), others indicated that the sex diff erence in performance arises from inherently diff erent biological capacities and women would not be able to reach men's levels of performance (Murphy, et al., 1986;Ford, et al., 2000;Cheuvront, et al., 2005). For longer ultra-distances, the sex diff erence in performance even seemed to increase. When the world best times for ultra-triathlon distances for both women and men were compared, men were ~19% faster than women in both a Double Iron and Triple Iron ultratriathlon (11.6 km swimming, 540 km cycling, and 126.6 km running), and ~30% faster in a Deca Iron ultra-triathlon (38 km swimming, 1,800 km cycling, and 422 km running; . Knechtle, Knechtle, and Lepers (2012) argued that the increase in sex diff erence with increasing length of an ultra-endurance performance such as an ultra-triathlon was most probably due to the lower skeletal muscle mass in women.
Long-distance triathlons such as the Ironman distance (3.8 km swimming, 180 km cycling, and 42.195 km running) and the Double Iron ultratriathlon (7.6 km swimming, 360 km cycling, and 84.4 km running) are examples of ultra-endurance events, defi ned as an event lasting more than six hours (Zaryski, & Smith, 2005). The Ironman triathlon, which started in 1978 (Lepers, 2008), and the Double Iron ultra-triathlon with the fi rst race in 1985 (Lenherr, Knechtle, Rüst, Rosemann, & Lepers, 2012), are relatively young ultra-endurance disciplines. The world fastest Double Iron ultra-triathlon times were achieved in 2011 by Adrian Brennwald (Switzerland) in 19 hours 50 minutes and for women in 1994 by Tina Bischoff (USA) in 22 hours 07 minutes. However, the latter race was held downriver and the swim split is not comparable to other races. The second fastest race time for women was achieved in 1991 by Astrid Benöhr (23 hours 45 minutes); the corresponding sex diff erence in overall performance was 19.8% (Table 1). The fastest Ironman race times in the Ironman Hawaii were more recently achieved, as the current Ironman Hawaii course record for men was set in 2011 by Craig Alexander (Australia) at 8 hours 3 minutes 56 seconds. Chrissie Wellington (Great Britain) set the women's course record in 2009 at 8 hours 54 minutes 2 seconds. The corresponding sex diff erence was 10.5%. The higher skeletal muscle mass in men might be an important variable of performance for ultra-endurance athletes and may explain the higher sex diff erence in the Double Iron ultra-triathlon performance compared to the Ironman distance. During an ultra-endurance performance, skeletal muscle mass decreased in men (Knechtle, Duff , Amtmann, & Kohler, 2008;Knechtle, Duff , Schulze, & Kohler, 2008; but not in women (Knechtle, Wirth, Baumann, Knechtle, Kohler, Rosemann, et al., 2010). The higher muscle mass may help men to maintain a faster speed during a longer time compared to women. When women and men were compared regarding their performance in cycling, the main factor accounting for sex diff erences in peak and mean power output during cycling was the muscle mass of the lower extremities (Perez-Gomez, Rodriguez, Ara, Olmedillas, Chavarren, González-Henriquez, et al., 2008). Women with a lower lean leg volume developed a lower peak power compared to men (Martin, Dore, Twisk, van Praagh, Hautier, & Bedu, 2004). The lower body fat in men might enhance endurance performance. Body fat has been shown as an important predictor variable for male Ironman triathletes . However, high body fat might enhance female ultra-endurance swim performance due to an increase in buoyancy.
A few studies investigated changes in sex diff erence in multi sports ultra-endurance events over time (Lepers, 2008;Rüst, Knechtle, Knechtle, Pfeifer, Rosemann, Lepers, et al., 2012). For long-distance triathlons, such as the Ironman triathlon, Lepers (2008) reported the sex diff erence in the top ten overall triathletes competing in Ironman Hawaii between 1988 and2007. Over this period, the mean sex diff erences in times for swimming, cycling, running, and total event were 9.8% (SD = 2.9), 12.7% (SD = 2.0), 13.3% (SD = 3.1), and 12.6% (SD = 1.3), respectively. Between 1988 and 2007, the sex diff erence remained identical for swimming (+0.1% per decade), increased for cycling (+0.8% per decade), and decreased for running (-2.8% per decade). The sex difference in overall race time also remained stable in the last two decades (-0.5% per decade). Regarding swimming as a separate discipline in triathlon, an actual study found no sex diff erence in ultra-swimming performance (Eichenberger, Knechtle, Knechtle, Rüst, Rosemann, & Lepers, 2012b). That women store more fat in the hips/buttocks than men has been used to suggest that women produce less drag in the water, thereby saving energy, which translates into potentially superior swimming performance during ultra-endurance swimming. Women also preferentially burn fat due to a sex-hormone advantage, perhaps making it possible for women to outperform men in ultra-endurance exercise (Dasilva, Guidetti, Buzzachera, Elsangedy, Krinski, De Campos, et al., 2011).
In this context, the aim of the present study was to compare the sex diff erences in swimming, cycling, and running performance and in overall race times between elite Ironman triathletes fi nishing Ironman Hawaii and Double Iron ultra-triathletes. To achieve this aim, the split and overall race times of the fastest female and male Ironman triathletes in Ironman Hawaii and Double-Iron ultra-triathletes during the period from 1999 to 2011 were analyzed. It was hypothesized that the sex diff erence in performance would be higher in the overall performance for Double Iron ultra-triathlon distance but lower for the swimming split compared to the Ironman triathlon.

METHOD
The split and the overall race times of female and male winners and top three overall fi nishers in Ironman Hawaii and Double Iron ultra-triathlons held worldwide between 1999 and 2011 were analyzed. The study was approved by the Institutional Review Board of St. Gallen, Switzerland, with waiver of the requirement for informed consent given that the study involved the analysis of publicly available data. For Double Iron ultra-triathlons, the race results were provided by the race directors. The corresponding author collected all race results personally over the last 20 years. The data set with the race results from Ironman Hawaii was downloaded from the race website. 3 In total, data were available from 20,638 athletes, including 5,163 women and 15,475 men in Ironman Hawaii and from 143 women and 1,252 men in Double Iron ultra-triathlon races held worldwide. For Ironman Hawaii and the Double Iron ultra-triathlons, data from professional and amateur athletes were combined in one group. Table 2 presents the races held in Double Iron ultra-triathlons between 1999 and 2011. Due to the low number of women participating in a Double Iron ultra-triathlon , races held before 1999 had to be excluded from analysis. From all other races, held between 1999 and 2011, overall race times and split times in swimming, cycling, and running of the winner (e.g., the fastest total race time) and the top three (e.g., the three fastest total race times) triathletes overall per sex and year were defi ned by Ironman Hawaii and in any participants in a Double Iron ultra-triathlon held worldwide. Split times and overall race times were converted to minutes. Afterwards, race results were analyzed regarding development of performance per split discipline and diff erences in performances between women and men. The sex diff erence was calculated as ([performance in men] -[performance in women]) / [performance in men] x 100). To facilitate reading, sex diff erences were converted to absolute values. The change in sex diff erence over time was also evaluated.

Statistical Analysis
To increase the reliability of data analyses, each set of data was tested for normal distribution as well as for homogeneity of variances before any other statistical analysis. The test for a normal distribution was performed using a D'Agostino-Pearson omnibus normality test and homogeneity of variances was tested using a Levene's test in case of two groups and with a Bartlett's test in case of more than two groups. To fi nd statistically significant changes in performances and sex diff erences over time, linear regressions were used. To fi nd diff erences between two groups, a Student's t test was used when the data was normally distributed and a Mann-Whitney test was used otherwise. To fi nd interactions between years and distance in sex diff erences, a two-way analysis of variance (ANOVA; time × distance) with subsequent Bonferroni post hoc analysis was performed. Some of the competitors participated for several years and were able to fi nish several times within the top three. To account for potential clustering for repeated athletes, we performed multi-level hierarchical regression analyses accounting for repeated measurements in athletes. Statistical analyses were performed using SPSS Version 19 (IBM SPSS, Chicago, IL, USA) and GraphPad Prism (Version 5, GraphPad Software, La Jolla, CA, USA). Signifi cance was accepted at p < .05 (two-sided for t tests).
For overall race times, the sex diff erence for the annual winners remained unchanged over time (p > .05) in Ironman Hawaii (r = .09) and Double Iron ultra-triathlons (r = .28). The sex diff erence for overall race time (Panel A) was 14.4% higher in Double Iron ultra-triathlons with 25.2% (SD = 10.4) for ultra-triathletes compared to 10.8% (SD = 2.3) for Ironman Hawaii triathletes (p < .05). For swimming, the sex diff erence for the annual winners remained unchanged in Ironman Hawaii (r = .13) and Double Iron ultra-triathlons (r = .15) over time (p > .05). The sex diff erence was 17.3% higher in Double Iron ultra-triathletes with 22.5% (SD = 10.0) compared to 5.2% (SD = 4.6) for Ironman Hawaii triathletes (p < .05). For cycling, the sex diff erence for the annual winners remained unchanged in Ironman Hawaii (r = .29) and Double Iron ultra-triathlon (r = .01) over time (p > .05). The sex diff erence was 11.7% higher in Double Iron ultra-triathlons with 23.7% (SD = 9.7) for ultra-triathletes compared to 12.0% (SD = 3.9) for Ironman Hawaii triathletes (p < .05). For running, the sex diff erence for the annual winners decreased in Ironman Hawaii (r = .46, p = .01) but increased in Double Iron ultra-triathlons (r = .39, p = .02) over time. In 2011, the sex diff erence was 5.3% higher in Double Iron ultratriathletes with 32.8% (SD = 15.6) compared to 27.5% (SD = 13.0) for Ironman Hawaii triathletes (p < .05).
The mean in overall race times and split times of the top three overall fi nishers and the annual winners with the corresponding sex diff erences for Ironman Hawaii and Double Iron ultra-triathlons are presented in Tables 3 and 4, respectively. For the top three fi nishers (Table 3), women were signifi cantly slower than men for all split times and overall race times. For winners in Ironman Hawaii (Table 4), men were faster than women for overall race times and split times. For Double Iron ultra-triathletes, women and men had similar times in swimming.
Results of the multi-level hierarchical regression analyses are presented in Table 5. Women were slower in both Ironman Hawaii and Double Iron ultra-triathlons. Race year showed an infl uence on race time in Ironman Hawaii, but not in Double Iron ultra-triathlon.

DISCUSSION
This study was intended to compare the sex diff erences in swimming, cycling, running, and overall race times between elite Ironman Hawaii triathletes in the Ironman World Championship, Ironman Hawaii, and Double Iron ultra-triathletes. It was hypothesized that the sex diff erence in performance would be higher in the overall performance for Double Iron ultra-triathlon distance but lower for the swim split compared to the Ironman Hawaii triathlon. However, the opposite was found. The sex difference in performance between the annual top three Ironman Hawaii and Double Iron ultra-triathletes was 17.4% higher in Double Iron ultra-triathlon for overall race time and 25.7% higher in Double Iron ultra-triathletes for the swimming split. A lower sex diff erence between the distances was found for cycling (15.3%), but not for running (24.6%).
Regarding the performance across time in the single Ironman distance in Ironman Hawaii', both women and men improved overall race time and in the running split and men also improved in the cycling split. In the Note.-*Split times in women were statistically signifi cantly diff erent from split times in men (p < .05). † Values in Ironman sex diff erences statistically signifi cantly diff erent from sex differences in Double Iron ultra-triathlon (p < .05). Double Iron ultra-triathlon, overall race times and split times remained unchanged in both women and men's races over time. An interesting fi nding was that the sex diff erence in performance remained unchanged over time for both Ironman Hawaii triathletes and Double Iron ultra-triathletes. The sex diff erence in Ironman Hawaii for the top three athletes between 1999 and 2011 was 11.0% (SD = 0.7) over time, with a smaller sex diff erence in swimming at 5.5% (SD = 1.8) compared to cycling at 12.5% (SD = 1.4) and running at 19.9% (SD = 7.5). For the top three Double Iron ultra-triathletes, the sex diff erence over time in overall performance was 17.4% higher compared to the sex diff erence in Ironman Hawaii 4 . Also for swimming (+25.7%), cycling (+15.3%), and running (+24.6%), the sex difference was higher in Double Iron ultra-triathletes as compared to Ironman Hawaii triathletes.
The sex diff erences in performance between Ironman Hawaii and Double Iron ultra-triathlon were similar for overall race time (17.4%) and the cycling split time (15.3%), but higher for swimming (25.7%) and running (24.6%). For the longer triathlon distance, swimming and running seemed to be the crucial disciplines for women. These diff erences might be explained by the sex diff erences within the disciplines, the history of the triathlete, and the equipment used. The sex diff erence in performance between Ironman Hawaii and Double Iron ultra-triathlon distance might be explained by the single disciplines. Recent studies suggested that women were able to achieve similar swimming times for longer distances since similar performances in female and male ultra-swimmers in a 12-hour ultra-swim (Eichenberger, Knechtle, Knechtle, Rüst, Rosemann, & Lepers, 2012a) and in the Channel Crossing (Eichenberger, Knechtle, Knechtle, Rüst, Rosemann, & Lepers, 2012b) have been reported. However, when the annual fastest swimmers in the Channel Crossing were considered, the top three male swimmers in the English Channel were approximately 12% faster than females over the last 36 years (Fischer, Knechtle, Rüst, & Rosemann, 2012). Also in an open-water ultra-swim in a lake, the sex diff erence remained unchanged at ~11.5% (Eichenberger, Knechtle, Knechtle, Rüst, Rosemann, Lepers, & Senn, 2012c). For cycling, the sex diff erence in performance was ~11% when diff erent cycling distances were investigated (Schumacher, Mueller, & Keul, 2001). In ultrarunning, however, the sex diff erence in performance was ~20% in 161 km ultra-marathons (Hoff man, 2010).
The increase in sex diff erence might also be explained with the increasing length of the single disciplines. A recent study investigated the change in sex diff erence in performance for Triple Iron ultra-triath-letes between 1988 and 2011 for the annual fastest athletes . For the Triple Iron distance, women were able to narrow the gap in swimming, but not in cycling and running. Men improved their overall race time, whereas women became slower across years. The sex diff erence in overall race time increased from 10% to 42% (r = .63, p < .01). In swimming, neither top men nor top women improved (p >.05) but the sex diff erence decreased from 35% in 1992 to 28% in 2011 (p < .05). Top men became faster in cycling, whereas top women became slower across years. The sex diff erence increased from 12% to 40% (p < .05). For running, top men showed no change in performance across years whereas top women became slower. The gender diff erence increased from 10% in 1992 to 64% in 2011 (p < .05). An actual study on sex-related participation and performance trends in ultra-mountain biking showed contrary results (Gloor, Knechtle, Knechtl, Rüst, Haupt, Rosemann, et al., 2013). In the Swiss Bike Masters between 1994 and 2012, the participation in men decreased while the participation in women remained low at ~2.4%. Performances of the annual fastest women women improved while performances of the annual fastest men remained unchanged. Top racers at world class level may start in shorter races and avoid these races that debilitate them for Olympic Games and World Championships.
Another explanation for the high sex diff erence in swimming performance between Ironman Hawaii and Double Iron ultra-triathlon could be the fact that wetsuits are prohibited in Ironman Hawaii but allowed in the Double Iron ultra-triathlons. Wearing a wetsuit leads to a signifi cantly lower swimming cadence (-14%), a signifi cantly lower heart rate (-11%), and a signifi cantly lower lactate values (-47%) compared to swimming without a wetsuit (Delextrat, Bernard, Hausswirth, Vercruyssen, & Brisswalter, 2003). Additionally, cycling effi ciency was signifi cantly higher (+12.1%) after swimming with a wetsuit compared to swimming without a wetsuit (Delextrat, et al., 2003). In an Ironman triathlon where athletes can wear a wetsuit in the swim, athletes without background as a swimmer may profi t from wearing a wetsuit and achieve faster swim times. However, in Ironman Hawaii', athletes with swimming backgrounds may swim fast in open water independently of whether they wear a wetsuit or not. It has been shown that wearing a wet suit improves swim performance more in ineffi cient swimmers with low buoyancy when swimming at low speeds (Chatared, Senegas, Selles, Dreanot, & Geyssant, 1995).
A further explanation for the sex diff erence in swimming and running performance between Ironman Hawaii and Double Iron ultra-triathlon could be the history of the athlete. It has been reported that many triathletes have been competitive swimmers (O'Toole, & Douglas, 1995). For recreational Ironman triathletes competing in Ironman Lanzarote, how-ever, 28% of the athletes had running backgrounds, 14% had been swimmers, and 13% cyclists (Gulbin & Gaff ney, 1999). In ultra-triathlon, the cycling and the running split showed the strongest eff ect on overall race time. For male Triple Iron ultra-triathletes, a signifi cant correlation (p < .05) was observed between total race time and both running time (r = .87) and cycling time (r = .62). In contrast, no correlation (p >.05) was shown between swim split times and overall race times (Knechtle, & Kohler, 2009). Most probably, triathletes competing in longer distances than the Ironman triathlon might prepare more specifi cally and have backgrounds as cyclists or runners. When training characteristics between Ironman and Triple Iron ultra-triathletes were compared, Triple Iron ultra-triathletes relied more on training volume in cycling and running, whereas speed in cycling training was related to race time in Ironman triathletes .
The multi-level hierarchical regression analyses showed that the inclusion of athletes fi nishing repeatedly within the top three had an infl uence on Double Iron ultra-triathlon race times, but not on Ironman Hawaii race times. The annual fi eld in Ironman Hawaii with approximately 2,000 starters is considerably higher compared to the approximately annual 100 starters in Double Iron ultra-triathlons. The approximately 20 times larger fi eld in Ironman Hawaii lowers the possibility for athletes to fi nish several times across years within the top three compared to the considerably smaller fi eld in Double Iron ultra-triathletes where fi nishing within the top three several times across years seems more likely.

Limitations
This study is limited because potential predictor variables for an ultra-endurance performance were not measured. Weather would only potentially aff ect year-by-year comparisons. The drastically diff erent participation numbers is a limitation. Although the top fi nisher and top three fi nishers were compared, smaller numbers of competing women (1/9 to 1/3 of men) suggests a deeper talent pool from which to draw in Double Iron ultra-triathlon competitions in particular. Race times from Double Iron ultra-triathlons were obtained from diff erent races held in diff erent locations. In contrast to Ironman Hawaii, held at the same site each year, the inclusion of diff erent race sites in Double Iron ultra-triathlon might infl uence race times due to diff erent topography. Split and race times in Ironman Hawaii were achieved using electronic timekeeping, while in the ultra-triathlons race times were measured by hand. While transition times were excluded in Ironman Hawaii, transition times were added to split times in cycling and running in ultra-triathlons as a general rule. In Ironman Hawaii wet suits are prohibited while wet suits are allowed in Double Iron ultra-triathlons. Additionally, disk wheels are prohibited on bicycles in Ironman Hawaii, but allowed in Double Iron ultra-triathlons. These diff erences in equipment might also have had an infl uence on both split times and overall race times.

Conclusions
These fi ndings suggest that the sex diff erence in performance is greater for Double Iron ultra-triathlon compared to a single Ironman triathlon distance for each split discipline and overall race time. The sex diff erence in performance for each discipline and overall race time has remained stable during the period between 1999 and 2011. It seems unlikely that women will be able to close the sex gap in ultra-triathlon performance in the near future. Further studies should examine the sex diff erence for other ultra-endurance events longer than a Double Iron ultra-triathlon such as Quintuple Iron, Deca Iron, and Double Deca Iron ultra-triathlon. The sex diff erence in swim performance might narrow, and the overall performance might extend, in longer triathlon distances. Future studies need to investigate the background of elite Ironman triathletes and ultra-triathletes whether they had backgrounds in swimming, cycling, or running.