Content uploaded by Thomas Johannes Rosemann
Author content
All content in this area was uploaded by Thomas Johannes Rosemann on Mar 13, 2016
Content may be subject to copyright.
Women outperform men in ultra-distance swimming 1
2
Women outperform men in ultra-distance swimming - 3
The ‘Manhattan Island Marathon Swim’ from 1983 to 2013 4
5
Beat Knechtle 1,2, Thomas Rosemann 1, Romuald Lepers 3 , Christoph Alexander Rüst 1
6
7
1 Gesundheitszentrum St. Gallen, St. Gallen, Switzerland 8
2 Institute of General Practice and Health Services Research, University of Zürich, 9
Zürich, Switzerland 10
3 INSERM U1093, Faculty of Sport Sciences, University of Burgundy, Dijon, France
11
12
13
14
15
16 Original Investigation 17
Abstract word count 190 18
Text word count 2,922 19
Number of figures 5 20
Number of tables 5 21
22
23
24 25
26
27
28
29 30
31
32
33
34
35 36
Corresponding author 37
PD Dr. med. Beat Knechtle 38
Facharzt FMH für Allgemeinmedizin 39
Gesundheitszentrum St. Gallen 40
Vadianstrasse 26 41
9001 St. Gallen 42
Switzerland 43
Telefon +41 (0) 71 226 82 82 44
Telefax +41 (0) 71 226 82 72 45
e-mail: beat.knechtle@hispeed.ch 46
1
Abstract 47
Purpose: Recent studies suggested that women and men’s ultra-swim performances may be 48
similar for distances of ~35 km. The present study investigated both the gender difference and 49
the age of peak ultra-swim performance between 1983 and 2013 at the 46-km ‘Manhattan 50
Island Marathon Swim’ with water temperatures <20°C. Methods: Changes in race times and 51
gender difference in 551 male and 237 female finishers were investigated using linear, non-52
linear, and hierarchical multi-level regression analyses. Results: The top ten race times ever 53
were significantly (P<0.0001) lower for women (371±11 min) than for men (424±9 min). 54
Race times of the annual fastest and annual three fastest women and men did not differ 55
between genders and remained stable across years. The age of the annual three fastest 56
swimmer increased from 28±4 years (1983) to 38±6 years (2013) (r2=0.06, P=0.03) in women 57
and from 23±4 years (1984) to 42±8 years (2013) (r2=0.19, P<0.0001) in men. Conclusions: 58
The best women were ~12-14% faster than the best men in a 46-km open-water ultra-distance 59
race with temperatures <20°C. The maturity of ultra-distance swimmers has changed during 60
the last decades with the fastest swimmers becoming older across the years. 61
Key words: swimming, gender difference, extreme, record 62
63
64
65
66
67
68
2
Introduction 69
70
Open-water ultra-distance swimming is increasing in popularity. Participation in the 34-km 71
‘English Channel Swim’ increased exponentially in the last ten years for both women and 72
men.1 The number of swimmers at the ‘English Channel Swim’ increased between 1991-2000 73
and 2001-2010 by 171% for men and by 135% for women, respectively.2 Similarly, an 74
increase in the rate of participation has been observed over the past years at the 26-km open-75
water lake swimming in both the ‘Marathon Swim Lake Zurich’ 3 and the ’12-hour indoor 76
swim’ held in Zürich 4. 77
78
Gender differences in ultra-distance swim performance have been investigated in both indoor 79
pool swimmers 4 and open-water ultra-endurance swimmers 1,2,5,6 and were leading to 80
disparate findings. It has been shown that the performance of the annual fastest women and 81
men did not differ in 12-hour indoor ultra-endurance swimmers 4 or in open-water ultra-82
endurance swimmers competing in the 34-km ‘English Channel Swim’ 2. In contrast, at the 83
26-km ‘Marathon Swim Lake Zurich’, the annual fastest men were on average 11.5% faster 84
than the annual fastest women.3 In the 32-km ‘Traversée Internationale du Lac St-Jean’, the 85
gender difference decreased from ~14% in 1973 to ~4% in 2012.5 And in the FINA 86
(Féderation Internationale de Natation) 10-km open-water race races, the gender difference 87
remained unchanged at ~7% between 2008 and 2012.6 Women seemed to narrow the gap with 88
increasing race distance. The higher body fat in female ultra-endurance swimmers 7 may be of 89
advantage for ultra-swims especially in cold water because thicker skinfolds could allow them 90
to endure longer in cold water 8. 91
92
3
These disparate findings of gender difference in ultra-distance swimming performance might 93
be due to the length of the swims, the water temperature and the level of the athletes. In the 94
34-km ‘English Channel Swim’ where the water temperature varied between 15°C and 18°C, 95
the annual fastest swimming speed (men 0.84±0.18 m/s; women 0.89±0.20 m/s) did not differ 96
between genders.2 Similarly, in a 12-hour indoor pool swim where water temperature was 97
kept constant at ~28°C, the annual fastest swimming speed did not differ between women and 98
men (men 0.88±0.06 m/s; women 0.79±0.19 m/s).4 In contrast, in the 26-km ‘Marathon Swim 99
in Lake Zurich’ where water temperature varied between 16.2 °C and 25.9 °C across years, 100
the annual male winner’s swimming speed was greater compared to the female swimming 101
speed (men 1.09±0.10 m/s; women 0.97±0.07 m/s).3 In other terms, the difference between 102
the annual fastest women (636.7 min) and men (674.6 min) in the 34-km ‘English Channel 103
Swim’ was 37.9 min where women were 5.6% faster than men.2 However, in the ’12-hour 104
indoor swim’, the difference between the annual fastest women (34.4 km) and men (38.3 km) 105
was 3.9 km where men were 10.2% faster than women.4 And in the 26-km ‘Marathon Swim 106
Lake Zurich’, the difference between the annual fastest women (452 min) and men (403 min) 107
was 49 min where men were 12.1% faster than women.3 It is therefore very likely that women 108
might outperform men in open-water ultra-distance swimming in a distance longer than then 109
34 km in the ‘English Channel Swim’ and at temperatures <20 °C. 110
111
Interestingly, the water temperature was significantly and negatively associated with 112
swimming speed for the annual top three swimmers in the 26-km ‘Marathon Swim Lake 113
Zurich’, suggesting that the colder the water, the longer the race time. For swimmers not 114
placed in the top three, race time was not associated with water temperature.3 Considering the 115
fitness level of the athletes, recreational athletes were investigated in the 12-hour indoor pool 116
swim 4 and in the ‘English Channel Swim’ 2. The ‘English Channel Swim’ is in fact not a race 117
where swimmers have to cross the Channel as solo swimmers. In contrast, elite swimmers 118
4
were investigated in the ‘Traversée Internationale du Lac St-Jean’ 5 and in the FINA 10-km 119
competitions 6. 120
121
Age could also be an important factor in ultra-distance swim performance. In a 12-hour 122
indoor swim, the best performances were achieved by women and men between 30 and 50 123
years of age.4 At the 26-km‘Marathon Swim in Lake Zurich’, the mean age of both female 124
and male winners did not differ between women (27.7±8.2 years, mean±SD) and men 125
(26.8±9.5 years, mean±SD).3 The age of the annual winners in the 26-km‘Marathon Swim in 126
Lake Zurich’ increased across the years with a mean of 30.9±6.5 years (mean±SD) for women 127
and 32.0±6.5 years (mean±SD) for men.3 Interestingly, increasing age was associated with an 128
increased risk for not finishing the 26-km‘Marathon Swim in Lake Zurich’.3 However, it has 129
been recently shown that for other ultra-endurance events such as the ‘Ironman Hawaii’, the 130
age of the fastest athletes tended to increase over recent decades.9 131
132
The purposes of the present study were to investigate (i) the gender difference in performance 133
in open-water ultra-distance swimmers and (ii) the age of peak ultra-swimming performance 134
in elite athletes competing at the ‘Manhattan Island Marathon Swim’ during the period of 135
1983-2012 where participants have to cover the distance of ~46 km at water temperatures 136
varying between 16.5°C and 20°C. Based upon existing reports for recreational and elite 137
open-water ultra-distance swimmers, we hypothesized (i) that elite female swimmers 138
competing in an open-water ultra-distance swimming race longer than ~35 km with a water 139
temperature < 20°C would achieve a similar performance to male swimmers or possibly 140
outperform men; and (ii) the age of peak ultra-swimming performance would increase across 141
the years. 142
5
Methods 143
144
Ethics 145
All procedures used in the study met the ethical standards of the Swiss Academy of Medical 146
Sciences and were approved by the Institutional Review Board of Kanton St. Gallen, 147
Kantonsspital St. Gallen, Switzerland with a waiver of the requirement for informed consent 148
of the participants given the fact that the study involved the analysis of publicly available 149
data. 150
151
The race 152
The ‘Manhattan Island Marathon Swim‘ is an open-water ultra-endurance swimming event 153
covering a full counter-clockwise circumnavigation of the island of Manhattan, New York, 154
USA, with a total distance of 28.5 miles (45.87 km) (www.nycswim.org). It generally starts at 155
the beginning of June at 7:40 a.m. at Battery Park City - South Cove. The field is generally 156
limited to 40 solo swimmers. The participants have a time limit of 9:30 h:min, must be 19 157
years or older and are not allowed to wear a wetsuit. To participate in this race, participants 158
must fulfill the qualification criteria. First, all applicants must document their competency to 159
participate in ‘Manhattan Island Marathon Swim‘. This can be done either by completing a 160
similar event or race, or a non-event qualifying swim logged by an observer in a prescribed 161
water temperature of 61°F (16.1°C) or colder and four hours or longer in duration. Secondly, 162
solo swimmers must have at least one and no more than two individuals to serve as support 163
crew aboard their assigned escort boat for this event. Water temperature in this event is 164
generally < 20 °C. In June, the water temperature in New York is between 16 °C and 19 °C 165
6
(http://www.currentresults.com/Oceans/Temperature/new-york-average-water-166
temperature.php). 167
Methodology 168
All athletes who had ever participated in the ‘Manhattan Island Marathon Swim’ between 169
1983 and 2013 were analyzed regarding participation, performance and age. The data set for 170
this study was obtained from the race website www.nycswim.org. Data before 1983 were not 171
complete and deemed not reliable for analysis. All male and female solo swimmers were 172
considered for data analysis. The race times and the ages of the annual top (e.g. fastest annual 173
race time) and of the annual top three overall (e.g. fastest annual three race times) women and 174
men were determined and analyzed to identify both the peak performance in swimming and 175
the peak age in swimming performance. Due to the low number of annual successful finishers 176
it was not possible to analyze a higher amount of annual data. Gender difference was 177
calculated using the equation ([race time in women] – [race time in men]) / [race time in men] 178
× 100 where gender difference was calculated for every pair of equally placed athletes (e.g. 179
the winner between men and women, the 2nd place between men and women, etc.). 180
Performance of the overall fastest, the three fastest and the ten fastest women and men ever 181
were determined and compared for the 31-year period. 182
183
Statistical analysis 184
In order to increase the reliability of the data analyses, each set of data was tested for normal 185
distribution as well as for homogeneity between variances prior to statistical analyses. Normal 186
distribution was tested using a D’Agostino and Pearson omnibus normality test and 187
homogeneity of variances was tested using a Levene’s test.10 To find significant changes in a 188
variable (e.g. race time, age) across years, regression analysis was used. A hierarchical multi-189
7
level regression model was used to avoid the influence of a cluster-effect (i.e. when athletes 190
finished more than one event) on the results for the annual top or annual top three 191
competitors. Regression analyses of performance were corrected for age to prevent a 192
misinterpretation of ‘age-effect’ with ‘time-effect’. Since the change in gender difference in 193
endurance is assumed to be non-linear 11, we additionally calculated to the linear also the non-194
linear regression model. We compared the linear to the best-fit non-linear model using 195
Akaike’s Information Criteria (AIC) and F-test in order to show which model would be the 196
most appropriate to explain the trend of the data. The results of the regression analyses (i.e. 197
whether the trend was varying over time or not) were confirmed using ANOVA (analysis of 198
variance). The differences between age and performance of the annual top and the annual top 199
three men and women, and between the top three and top ten women and men ever where 200
investigated using a Student’s t-test with Welch’s correction in case of unequal variances for 201
normally distributed and a Mann-Whitney test for non-normally distributed data. Statistical 202
analyses were performed using SPSS Statistics (Version 21, IBM SPSS, Chicago, IL, USA) 203
and GraphPad Prism (Version 6.01, GraphPad Software, La Jolla, CA, USA). P<0.05 was 204
used to imply statistical significance (two-tailed for t-tests). Data in the text are reported as 205
mean ± standard deviation (SD). 206
207
208
209
210
211
212
213
8
Results 214
215
Participation and finisher trends 216
Between 1983 and 2013, a total of 909 swimmers (640 men and 269 women) started while 217
551 men and 237 women finished. On average, 8±4 women and 18±9 men finished the race 218
annually (Figure 1A). The annual number of female and male participants remained 219
unchanged across years. The overall rate of finishers was 86.4±18.0% (i.e. women 220
87.6±20.3% and men 85.8±18.6%) (Figure 1B). Among the total male finishers, 85 swimmers 221
finished the race at least twice. The lowest number of finishers was in 2005 where only two 222
men finished. The largest number of male finishes (n=17) belongs to one athlete, Kristian 223
Rutford from the United States of America. Among the female finishers, 44 swimmers 224
finished more than once and Shelley Taylor-Smith from Australia obtained the highest record 225
with eight successful finishes. 226
227
The best performances 228
The race record for women was achieved in 1995 by Shelley Taylor-Smith in 345 min. This 229
time is 14.1% lower than the race record for men set in 1985 by Drury Gallagher in 402 min 230
(Table 1). The three fastest performances ever did not differ statistically between women 231
(357±11 min) and men (413±11 min) (Table 1). However, when the ten fastest race times 232
were considered, women were significantly faster than men (P<0.0001) with a gender 233
difference in performance of 12.4±1.0% (Table 1). 234
235
236
9
Performance trends and gender difference in performance 237
The annual fastest women completed the race in 440.0±44.8 min, the differences in 238
swimming time across years did not change (r2=0.06, P=0.3757) (Figure 2), also when 239
controlled for athletes with multiple finishes (Table 2). The annual fastest race time for men 240
was 458.5±29.4 min and did not change across years (r2=0.04, P=0.131) (Figure 2), also when 241
controlled for multiple finishes (Table 2). The annual three fastest race times remained stable 242
across the years for both women (Figure 3A) (461.3±31.7 min, r2=0.02, P=0.1117) and men 243
(Figure 3B) (468.8±27.2 min, r2=0.002, P=0.0985) also when controlled for multiple finishes 244
(Table 2). ANOVA confirmed the linear trend in performance for the annual three fastest men 245
(r2=0.018, P=0.03) but not for the annual three fastest women (r2=0.00017, P=0.74) where the 246
change was non-linear (i.e. polynomial regression 3rd degree) (Table 3). The corresponding 247
gender difference in performance remained unchanged across years at 6.9±10.7% (r2=0.02, 248
P>0.05) for the annual fastest (Figure 4A) and at 4.5±3.9% (r2=0.04, P=0.3475) for the annual 249
three fastest swimmers (Figure 4B). ANOVA confirmed the linear trend for the change in 250
gender difference in the annual three fastest (r2=0.04, P=0.027) (Table 4). 251
252
Age trends over time 253
The annual fastest men (Figure 5A) became older across years from 27 years (1984) to 34 254
years (2013) (r2=0.22, P=0.008) also when controlled for athletes with multiple finishes 255
(Table 5). The age of the annual fastest women (Figure 5A) remained unchanged at 28.1±6.9 256
years (r2=0.01, P>0.05) also when controlled for athletes with multiple finishes (Table 5). 257
However, the annual three fastest women and men became older across years (Figure 5B) also 258
when controlled for athletes with multiple finishes (Table 5). In women, the age of the annual 259
three fastest increased from 28±4 years (1983) to 38±6 years (2013) (r2=0.06, P=0.03). In 260
men, the age of the annual three fastest increased from 23±4 years (1984) to 42±8 years 261
10
(2013) (r2=0.19, P<0.0001). ANOVA confirmed the linear trend in the change of age for the 262
annual three fastest men (r2=0.13, P=0.0002) and women (r2=0.07, P=0.012). 263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
11
Discussion 280
281
The first aim of this study was to investigate the gender difference in performances at the 46-282
km open-water ‘Manhattan Island Marathon Swim’ between 1983 and 2013. Interestingly, the 283
results showed that the best women outperformed the best men by ~12-14%. This finding 284
contradicts results from other ultra-endurance events such as ultra-running 12, ultra-cycling 285
13and ultra-triathlon 14 where men usually were faster than women. 286
287
Regarding the participation in the ‘Manhattan Island Marathon Swim’, the number of male 288
and female swimmers showed no change across the years and the percentage of finishers 289
remained stable. The unchanged number of participants is due to the fact that the field is 290
limited to 40 solo swimmers (www.nycswim.org). In contrast, in the ‘English Channel Swim’ 291
was an exponential increase in the number of participants in the last decades.1 However, the 292
‘English Channel Swim’ is not a race since each swimmer has to cross the Channel alone 293
followed by his support boat. A potential reason for the limited field in ultra-distance 294
swimming in an official race is logistical. The security of the swimmers and the limited 295
number of available support boats might be the most important reasons. 296
297
Between 1983 and 2013, both women and men showed no improvement in performance. This 298
is in line with 12-hour ultra-swimmers in a pool during the 1996-2010 period 4 and open-299
water ultra-swimmers in the ‘English Channel Swim’ between 1975 and 2011 1. These time 300
frames of ~30 years seemed to be too short to find an improvement in performance in contrast 301
to the 1955-2012 period in ‘La Traversee Internationale du Lac St-Jean’ where the fastest 302
women and men improved over time.5 An improvement in performance across years in open-303
12
water ultra-swimmers might be due to changes over time in anthropometric characteristics 304
such as body height. It has been shown that the world’s fastest 100 m swimmers became taller 305
and heavier between 1912 and 2008.15 For open-water ultra-swimmers, anthropometric 306
characteristics such as body height and body mass index were predictive for race time.16 307
308
A potential physiological explanation for the faster race times in women could be the higher 309
body fat in female ultra-distance swimmers. Recent studies reported a body fat percentage of 310
30.7±3.7% 7 to 31.3±3.6% 16 for female open water ultra-swimmers compared to 18.8±4.5% 7
311
to 20.2±5.6% 16 for male open water ultra-distance swimmers. Competitive female swimmers 312
have proportionately more fatty tissue at the lower body than male swimmers.17 The higher 313
percentage of body fat may improve both buoyancy and swimming performance in women. 314
The higher body fat may also improve women’s swimming performance in cold water 315
swimming by acting as insulation against the cold. A case report describing two male 316
swimmers in water at 4°C showed that the swimmer with more body fat (23.4%) was able to 317
complete 2.2 km within 42 min whereas the swimmer with lower body fat (21.0%) stopped 318
after 1.3 km.18 Keatinge et al. also reported that swimmers with less subcutaneous fat 319
terminated a swim in water at temperatures of 9.4 °C to 11.0 °C after significantly less time in 320
the water than those with thicker skinfold thickness.8 Swimmers with less thick subcutaneous 321
fat made significantly shorter swims than those with thicker fat layers. The thinnest subject 322
swam for only 23 min, whereas the four with the thickest fatty layers swam for over 60 min.8 323
Branningan et al. investigated 70 male and 39 female swimmers in a 19.2-km open water 324
swimming race in Perth, Western Australia.19 In the study by Branningan et al., hypothermia 325
defined as body core temperature of <35 °C, was the most common race-related illness.19 326
Longer race duration was also associated with an increased risk of hypothermia, and higher 327
body mass index was associated with a decreased risk of hypothermia.19 Taken together, their 328
13
data suggest that women with higher body fat may stay longer in cold water compared to men 329
with lower body fat. 330
331
Apart from gender differences in body fat, swimming efficiency is also different between 332
women and men. Buoyancy is higher in women through a lower ‘underwater torque’, which 333
can be defined loosely as the tendency for the feet to sink.20 In addition, and in contrast to 334
running where the energy cost appeared similar between women and men, the energy cost of 335
freestyle swimming has been shown to be significantly higher (i.e. lower economy) in men 336
compared to women.20,21 The energy cost of swimming is a valuable parameter to quantify 337
swimming economy. At a swim speed of 1 m/s, differences in drag force and coefficient of 338
drag have been reported between women and men.22 The energy cost of swimming depends 339
primarily on the propelling efficiency of the arm stroke and the hydrodynamic resistance. 340
However, it has been suggested that gender differences in energy costs of swimming were 341
mainly attributed to differences in hydrodynamic resistance.23 Regarding the influence of 342
anthropometry on swimming efficiency, women also have a smaller body size resulting in 343
smaller body drag, a smaller body density with a greater body fat percent and shorter lower 344
limbs, resulting overall in a more horizontal and streamlined position and therefore a smaller 345
underwater torque.20,24 346
347
Apart of anthropometric characteristics, swimming economy should also be considered as an 348
explanation for the gender difference. Swimming economy is considered as one of the most 349
important predictors in swimming performance.25-27 There are three swimming economy 350
related parameters known such as the net energy cost corresponding to v VO2max (Cv 351
VO2max), the slope of the regression line obtained from the energy expenditure (E) and 352
corresponding velocities during an incremental test (Cslope) and the ratio between the mean E 353
14
value and the velocity mean value of the incremental test (Cinc).28 The investigation of the 354
relationship between the time limit at the minimum velocity that elicits the individual's 355
maximal oxygen consumption (TLim-v VO2max) and the above mentioned swimming 356
economy related parameters showed that TLim-v VO2max seemed to depend in women more 357
on swimming economy than in men.28 358
359
Another explanation for the performance in women might be drafting during open-water 360
swimming. In triathlon 29 and in open-water ultra-distance swimming 5, athletes draft one 361
behind the other. During the ‘Manhattan Island Marathon Swim’, women may draft behind 362
men and reduce drag.30 Drafting may save energy since swimming behind another swimmers 363
reduced oxygen uptake, heart rate, blood lactate and stroke rate. For the last part of the race, 364
the best women may have enough energy to pass and leave the leading men. 365
366
In the present study, the ten best women ever were ~12-14% faster than the ten best men ever 367
when the fastest race times ever were analyzed and the mean gender difference in 368
performance across years was ~5-7%. When the three best women and men ever were 369
compared, the performances did not differ between women and men. Any difference might 370
not have been identified because of the small sample size. When taken as the whole cohort, 371
the gender difference in ultra-endurance performance appears higher and men were faster than 372
women. For example, in ultra-endurance cyclists competing in the ‘Race Across America’ 373
between 1982-2012, the fastest men were 14-15% faster than the fastest women and the 374
gender difference was ~25% for the annual three fastest women and men in the last 30 375
years.13 In running, the gender difference is at ~11-12% when considering running distances 376
from 100m to 200km.31,32 In 161-km trail running, the gender difference was even at ~20%.12 377
According to Cheuvront et al., the gender difference in running performance appears 378
15
biological in origin.33 Success in distance running and sprinting is determined largely by 379
aerobic capacity and muscular strength with men having a larger aerobic capacity and greater 380
muscular strength, respectively.33 Therefore, the gap in running performance between women 381
and men is unlikely to narrow naturally. This might be true for running and ultra-running but 382
our results suggest not for ultra-swimming in cold water. 383
384
The second aim of this study was to investigate the change in the age of peak ultra-swimming 385
performance across years. Based upon recent findings for long-distance triathletes it was 386
hypothesized that the age of the fastest swimmers would increase across years. Indeed, the 387
age of the annual fastest men increased over time whereas the age of the annual fastest 388
women remained unchanged. For both the annual three fastest women and men the age of the 389
fastest race times increased across years also when controlled for athletes with multiple 390
finishes. In 2012, the annual three fastest women and men were older than 35 years. By 391
definition, these were master athletes defined as athletes older than 35 years and 392
systematically training for and involved in organized forms of sport specifically designed for 393
athletes older than 35 years.34 However, previous studies suggested that athletes in ultra-394
endurance races became older across years without an impairment of performance. For 395
example, in the Ironman World Championship ‘Ironman Hawaii’ the annual top ten finishers 396
became older and faster across years.9 The annual ten fastest Ironman triathletes in ‘Ironman 397
Hawaii’ were at the age of ~35 years for both women and men.9 In ultra-marathon runners 398
competing in ‘Spartathlon’ and ‘Badwater’ as two of the toughest ultra-marathons held 399
worldwide, the fastest runners were 40-45 years old.35 It seems that age, as a performance 400
limiting factor, in ultra-endurance moved to higher ages in recent years. 401
402
403
16
Practical Applications 404
The results of the present study suggest that women may outperform men in ultra-distance 405
swimming, especially in cold water. These last three decades at the 46-km ‘Manhattan Island 406
Marathon Swim’, the best women outperformed the best men by ~12-14% while the annual 407
best performance remained stable for both women and men across years. The age of the 408
annual fastest male swimmers became older over time. The maturity of these ultra-distance 409
swimmers changed during the last decades where the fastest swimmers became older over 410
time. Future studies need to compare anthropometric, training and physiological variables of 411
the fastest open-water ultra-distance swimmers. Specifically, body core temperature should be 412
recorded in the fastest open-water ultra-distance swimmers and correlated to their body fat 413
and body mass index. 414
415
Conclusions 416
The best women were ~12-14% faster than the best men in the 46-km open-water ultra-417
distance swimming ‘Manhattan Island Marathon Swim’ held at temperatures <20°C. The 418
annual fastest women were faster than the annual fastest men and it seems unlikely that men 419
would be able to overtake women in this specific race since the changes were linear, but not 420
non-linear, across years. The maturity of ultra-distance swimmers has changed during the last 421
years with the fastest swimmers becoming older across the years. 422
423
424
425
426
427
17
References 428
429
1 Fischer G, Knechtle B, Rüst CA, Rosemann T. Male swimmers cross the English 430
Channel faster than female swimmers. Scand J Med Sci Sports. 2013;23:e48-e55. 431
432
2 Eichenberger E, Knechtle B, Knechtle P, Rüst CA, Rosemann T, Lepers R. Best 433
performances by men and women open-water swimmers during the 'English Channel 434
Swim' from 1900 to 2010. J Sports Sci. 2012;30:1295-1301. 435
436
3 Eichenberger E, Knechtle B, Knechtle P, Rüst CA, Rosemann T, Lepers R, Senn O. 437
Sex difference in open-water ultra-swim performance in the longest freshwater lake 438
swim in Europe. J Strength Cond Res. 2013;27:1362-1369. 439
440
4 Eichenberger E, Knechtle B, Christoph AR, Knechtle P, Lepers R, Rosemann T. No 441
gender difference in peak performance in ultra-endurance swimming performance - 442
analysis of the 'Zurich 12-h Swim' from 1996 to 2010. Chin J Physiol. 2012;55:346-443
351. 444
445
5 Rüst CA, Knechtle B, Rosemann T, Lepers R. Women reduced the sex difference in 446
open-water ultra-distance swimming – ‘La Traversée Internationale du Lac St-Jean’ 447
1955-2012. Appl Physiol Nutr Metab, 2014;39:270-273. 448
449
6 Vogt P, Rüst C, Rosemann T, Lepers R, Knechtle B. Analysis of 10 km swimming 450
performance of elite male and female open-water swimmers. SpringerPlus 2013; 451
2:603 452
453
7 Weitkunat T, Knechtle B, Knechtle P, Rüst CA, Rosemann T. Body composition and 454
hydration status changes in male and female open-water swimmers during an ultra-455
endurance event. J Sports Sci. 2012;30:1003-1013. 456
457
8 Keatinge WR, Khartchenko M, Lando N, Lioutov V. Hypothermia during sports 458
swimming in water below 11 degrees C. Br J Sports Med. 2001;35:352-353. 459
460
9 Gallmann D, Knechtle B, Rüst CA, Rosemann T, Lepers R. Elite triathletes in 461
‘Ironman Hawaii’ get older but faster. Age (Dordr), 2014;36:407-416. 462
463
10 D'Agostino RB. ‘Tests for Normal Distribution’ in Goodness-Of-Fit Techniques, 464
edited by D'Agostino RB and Stepenes MA, Macel Decker, 1986. 465
466
11 Reinboud W. Linear models can't keep up with sport gender gap. Nature. 467
2004;432(7014):147 468
469
12 Hoffman MD. Performance trends in 161-km ultramarathons. Int J Sports Med. 470
2010;31:31-37. 471
472
13 Rüst CA, Knechtle B, Rosemann T, Lepers R. Men cross America faster than women - 473
the 'Race Across America' (RAAM) from 1982 to 2012. Int J Sports Physiol Perform. 474
2013;8:611-617. 475
18
14 Rüst CA, Knechtle B, Rosemann T, Lepers R. Sex difference in race performance and 476
age of peak performance in the Ironman Triathlon World Championship from 1983 to 477
2012. Extrem Physiol Med. 2012;1:15. 478
479
15 Charles JD, Bejan A. The evolution of speed, size and shape in modern athletics. J Exp 480
Biol. 2009;212:2419-2425. 481
482
16 Knechtle B, Baumann B, Knechtle P, Rosemann T. Speed during training and 483
anthropometric measures in relation to race performance by male and female open-484
water ultra-endurance swimmers. Percept Mot Skills. 2010;111:463-474. 485
486
17 McLean SP, Hinrichs RN. Sex differences in the centre of buoyancy location of 487
competitive swimmers. J Sports Sci. 1998;16:373-383. 488
489
18 Knechtle B, Christinger N, Kohler G, Knechtle P, Rosemann T. Swimming in ice cold 490
water. Ir J Med Sci. 2009;178:507-511. 491
492
19 Brannigan D, Rogers IR, Jacobs I, Montgomery A, Williams A, Khangure N. 493
Hypothermia is a significant medical risk of mass participation long-distance open 494
water swimming. Wilderness Environ Med. 2009;20:14-18. 495
496
20 Pendergast DR, Di Prampero PE, Craig AB, Wilson DR, Rennie DW. Quantitative 497
analysis of the front crawl in men and women. J Appl Physiol. 1977;43:475-479. 498
499
21 Pate RR, Sparling PB, Wilson GE, Cureton KJ, Miller BJ. Cardiorespiratory and 500
metabolic responses to submaximal and maximal exercise in elite women distance 501
runners. Int J Sports Med. 1987;8:91-95. 502
503
22 Zamparo P. Effects of age and gender on the propelling efficiency of the arm stroke. 504
Eur J Appl Physiol. 2006;97:52-58. 505
506
23 Toussaint HM, de Groot G, Savelberg HH, Vervoorn K, Hollander AP, van Ingen 507
Schenau GJ. Active drag related to velocity in male and female swimmers. J 508
Biomech. 1988;21:435-438. 509
510
24 Lavoie JM, Montpetit RR. Applied physiology of swimming. Sports Med. 1986;3:165-511
189. 512
513
25 Fernandes RJ, Vilas-Boas JP. Time to exhaustion at the VO2max velocity in 514
swimming: A review. J Hum Kinet. 2012;32:121-134. 515
516
26 Chatard JC, Lavoie JM, Lacour JR. Analysis of determinants of swimming economy 517
in front crawl. Eur J Appl Physiol Occup Physiol. 1990;61:88-92. 518
519
27 Smith DJ, Norris SR, Hogg JM. Performance evaluation of swimmers: scientific tools. 520
Sports Med. 2002;32:539-554. 521
522
28 Fernandes RJ, Billat VL, Cruz AC, Colaço PJ, Cardoso CS, Vilas-Boas JP. Has gender 523
any effect on the relationship between time limit at VO2max velocity and swimming 524
economy? J Hum Movement Stud. 2005;49:127-148 525
19
29 Landers GJ, Blanksby BA, Ackland TR, Monson R. Swim positioning and its 526
influence on triathlon outcome. Int J Exerc Sci. 2008;1:96-105. 527
528
30 Chatard JC, Wilson B. Drafting distance in swimming. Med Sci Sports Exerc. 529
2003;35:1176-1181. 530
531
31 Coast JR, Blevins JS, Wilson BA. Do gender differences in running performance 532
disappear with distance? Can J Appl Physiol. 2004;29:139-145. 533
534
32 Sparling PB, O'Donnell EM, Snow TK. The gender difference in distance running 535
performance has plateaued: an analysis of world rankings from 1980 to 1996. Med Sci 536
Sports Exerc. 1998;30:1725-1729. 537
538
33 Cheuvront SN, Carter R, Deruisseau KC, Moffatt RJ. Running performance 539
differences between men and women: an update. Sports Med 2005;35: 1017-1024. 540
541
34 Reaburn P, Dascombe B. Endurance performance in masters athletes. Eur Rev Aging 542
Phys Act. 2008;5:31-42. 543
544
35 Zingg MA, Knechtle B, Rüst CA, Rosemann T, Lepers R. Analysis of participation 545
and performance in athletes by age group in ultramarathons of more than 200 km in 546
length. Int J Gen Med. 2013;6:209-220. 547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
20
Record time (min) for men in1985
402
Record time (min) for women in 1995
345
Gender difference in performance (%)
14.1
Top three race time (min) for men
413±11
Top three race time (min) for women
357±11
Gender difference in performance (%)
13.6±0.4
Top ten race time (min) for men
424±9
Top ten race time (min) for women
371±11 ***
Gender difference in performance (%)
12.4±1.0
572
Table 1: Record and performance of the top three and top ten fastest finishers during the 573
1983-2012 period with corresponding gender difference in performance. Results are 574
expressed as mean±SD. *** = P<0.0001 575
576
577
578
579
580
581
582
583
584
21
Model
ß
SE (ß)
Stand. ß
T
P
Annual fastest men
1
0.638
0.619
0.191
1.031
0.311
2
0.638
0.619
0.191
1.031
0.311
3
1.190
0.681
0.357
1.749
0.092
Annual three fastest men
1
0.154
0.367
0.046
0.421
0.675
2
0.154
0.367
0.046
0.421
0.675
3
-0.033
0.407
-0.010
-0.081
0.936
Annual fastest women
1
1.262
0.944
0.249
1.338
0.192
2
1.262
0.944
0.249
1.338
0.192
3
1.261
0.968
0.249
1.302
0.204
Annual three fastest women
1
0.623
0.508
0.137
1.225
0.224
2
0.623
0.508
0.137
1.225
0.224
3
0.465
0.523
0.102
0.889
0.377
585
Table 2: Multi-level regression analyses for change in performance across years for women 586
and men (Model 1) with correction for multiple finishes (Model 2) and age of athletes with 587
multiple finishes (Model 3) 588
22
Swimming speed
Kind of
regression
Sum of
Squares
DOF
AICC
Best
regression
AIC-Test
Best
regression
F-Test
Delta
Probability
Likelihood
Annual fastest men
polynomial
22394.1
26
213.59
linear linear 3.90 0.12 87.57%
linear
25066.6
29
209.69
Annual fastest women
polynomial
49449.8
26
229.14
linear linear 2.87 0.19 80.81%
linear
52689.4
28
226.27
Annual three fastest men
polynomial
19603.8
18
231.13
linear linear 31.90 1.17 e-07 100%
linear
21391.5
28
199.22
Annual three fastest women
polynomial
18282.3
24
188.48
polynomial polynomial 4.29 0.10 89.53%
linear
25339
26
192.77
Table 3: Comparison of linear and non-linear regression analyses of changes in swimming speed across years for women and men to determine
which model is the best
589
590
591
592
593
23
Gender difference
Kind of
regression
Sum of
Squares
DOF
AICC
Best
regression
AIC-Test
Best
regression
F-Test
Delta
Probability
Likelihood
Annual fastest
polynomial
2871.75
24
149.34
linear linear 7.92 0.018 98.13%
linear
3113.91
28
141.41
Annual three fastest
polynomial
209.24
19
79.89
linear linear 4.85 0.081 91.87%
linear
378.19
26
75.04
Table 4: Comparison of linear and non-linear regression analyses of changes in gender difference across years for women and men to determine
which model is the best
24
Model
ß
SE (ß)
Stand. ß
T
P
Annual fastest men
1
0.346
0.121
0.476
2.863
0.008
2
0.346
0.121
0.476
2.863
0.008
Annual fastest three men
1
0.414
0.093
0.435
4.460
< 0.001
2
0.414
0.093
0.435
4.460
< 0.001
Annual fastest women
1
0.093
0.149
0.119
0.623
0.539
2
0.093
0.149
0.119
0.623
0.539
Annual fastest three women
1
0.234
0.103
0.248
2.271
0.026
2
0.234
0.103
0.248
2.271
0.026
594
Table 5: Multi-level regression analyses for change in age across years for women and men 595
(Model 1) and with correction for multiple finishes (Model 2) 596
597
598
599
600
601
602
603
25
Figure captions 604
605
Figure 1 Number of female, male and overall participants (Panel A) and percent finisher rate 606
for men, women and overall (Panel B) across years 607
608
Figure 2 Race times of the annual fastest women and men across years 609
610
Figure 3 Race times of the annual top three women (Panel A) and men (Panel B) across 611
years. Results are presented as mean ± SD 612
613
Figure 4 Gender difference in performance of the annual fastest (Panel A) and the annual 614
three fastest (Panel B) swimmers across years. Results are presented as mean ± SD for the 615
annual three fastest 616
617
Figure 5 Age of the annual top (Panel A) and the annual top three (Panel B) men and women 618
across years. Results are presented as mean ± SD for the annual top three 619
620
621
622
623
26
624
Figure 1 625
27
Swim Tim e (m in)
626 627
Figure 2 628
28
629
Figure 3 630
631
632
29
633 Figure 4 634
30
635 Figure 5 636
637
31
