*2.1. Participants*

Fifteen recreational marathon runners (age: 42 ± 7 years, height: 174.9 ± 6.5 cm and body mass: 72.8 ± 6.9 kg) volunteered to take part in the study. All participants were healthy and ran approximately 1–2 years on a systematic basis with a structured program with an average weekly load of 50–60 km. Based on their finishing time at the Athens Marathon 2018, athletes were divided into a moderate-level group, with finishing times < 240 min (*n* = 8), and a low-level group with finishing times > 240 min (*n* = 7). Before the start of the study, the institutional review board committee approved the experimental protocol in accordance with the Helsinki Declaration.

#### *2.2. Maximal Incremental Test*

Three to nine days before participation in the Athens Marathon 2018 race, participants performed a maximal incremental test on a treadmill (Technogym run race 1200, Italy) for the determination of VO2 max, maximum aerobic velocity (MAV), maximum heart rate (HRmax), the relationship between blood lactate concentration and running velocity, oxygen consumption and running velocity, heart rate and running velocity, and the energy cost of running.

The protocol started at 7 km·h−<sup>1</sup> and was increased by 1.5 km·h−<sup>1</sup> every 3 min until volitional exhaustion. Treadmill grade was set at 1% throughout the protocol. Gas exchange was measured by the open circuit Douglas bag method as described by Cooke (2009). The subject breathed through a low-resistance 2-way Hans-Rudolph 2700 B valve (Shawnee, OK, USA). The concentrations of CO2 and O2 in the expired air were measured by using the Hi-tech (GIR 250) combined Oxygen and Carbon Dioxide Analyzer. The gas analyzers were calibrated continuously against standardized gases (15.35% O2, 5.08% CO2 and 100% N2). Expired volume was measured by means of a dry gas meter (Harvard) previously calibrated against standard air flow with a 3 L syringe. Barometric pressure and gas temperature were recorded and respiratory gas exchange data for each work load (i.e., VO2, VCO2 and VE) were determined based on the computations described by Cooke [23] when VEatps, FECO2 and FEO2 are known. The highest VO2 value obtained during a 30-sec time period during the incremental exercise test was recorded as the subject's .VO2max. HR was continuously measured telemetrically (Polar RS400) and the highest 10 sec value was regarded as maximal. The test was considered as maximal when at least 3 of the following criteria were achieved: (a) visual exhaustion of the participants, (b) a plateau in oxygen consumption (<2 mL kg−1·min−1) despite an increase in running velocity, (c) maximal HR higher than 90% of the predicted maximum (220-age) and (d) maximum respiratory exchange ratio > 1.1. MAV was calculated using the following formula: MAV (km·h−1) = Velocity of the last completed stage + (seconds run at last stage/180).

#### *2.3. LTh and LTP Determination*

At the end of each stage during the incremental test, approximately 0.3 μL of whole blood was collected from the fingertip and immediately analyzed for lactate concentration with a portable analyzer (Lactate Pro 2, Arkray Factory Inc., Koka-Shi, Japan) using an enzymatic-amperometric method. The individual relationships between blood lactate concentrations and running velocities were determined using an exponential model: y = a + b × exp(x/c), where y = lactate concentration, x = running velocity and a, b and c are constants. The LTh and the LTP were identified as the velocities (km·h−1) at which blood lactate concentrations were increased by 0.3 and 1.5 mmol·L−<sup>1</sup> from baseline values, respectively. Furthermore, LTh and LTP were expressed relative to MAV (%MAV) units, and based on the relationship between VO2 and running velocity were also expressed in absolute (mL·kg−1·min−1) and relative (%VO2 max) VO2 max values.

#### *2.4. Energy Cost of Running*

The gas exchange data (VO2, VCO2) collected during the final 30 s of every 3-min stage up to the previous stage from the LTP were used for the calculations of the caloric cost of running. Substrate oxidation rate (g·min−1) was estimated using nonprotein respiratory quotient equations [24]:

Fat oxidation (g·min−1) = 1.6947 × VO2 (L·min−1) − 1.7012 × VCO2 (L·min−1)

Carbohydrate oxidation (g·min−1) = 4.5851 × VCO2 (L·min−1) − 3.22259 × VO2 (L·min−1)

The energy produced from each substrate was calculated by assuming an energy equivalent for 1 g of fat and carbohydrate of 9.75 and 4.07 kcal, respectively [25]. Total ECr was quantified from the sum of these values and was expressed in kcal·kg−1·km−1. The energy cost of running at 10 km/h and at the velocities corresponding to LTh and marathon race pace were estimated from the relationship between exergy cost and running velocity derived from the incremental test.

All the physiological data were analyzed after the completion of the marathon race to avoid any pacing strategies from the participants and their coaches based on the results of testing.

#### *2.5. Route Characteristics and Race Pace Analysis*

The profile of the Athens Marathon route includes rolling uphills and downhills. More specifically, when calculated in 450 m intervals, the total ascent, total descent, the percent of uphill distance, the percent of downhill distance and the steepest uphill and downhill, respectively, are for: (a) the total route: 317 m, 262 m, 51.2%, 40.5%, 3.8% and −6.2%, (b) the 0–10 km split: 19 m, 36 m, 36.4%, 50%, 1.3% and −2.0%, (c) the 10–21.1 km split: 143 m, 66 m, 66.7%, 25%, 3.6% and −6.2%, (d) the 21.1–30 km split: 122 m, 5 m, 95%, 5%, 3.3% and −1.1% and (e) the 30–42.195 km split: 32 m, 155 m, 15.4%, 76.9%, 3.8% and −5.1% [22].

Finishing time and split times for each participant were exported from the official results posted on the site of the organization [21]. The average running velocity of each runner was calculated by dividing marathon distance to the time needed to complete the race. Race pace was expressed as a percentage of VO2 max (index of fractional utilization of VO2 max), MAV and the velocities at LTh (vLTh) and LTP (vLTP).

To determine the differences between the two groups in pacing during the race, average running velocities for the distances of 0–10, 10–21.1, 21.1–30 and 30–42.195 km were calculated by dividing the distance of the split to the time to complete the split. For the analysis of the data, mean velocity of each split was expressed as a percentage of the vLTh. The velocity at LTh was selected as a reference point because for the whole sample, average marathon running velocity was equal to vLTh.

## *2.6. Statistical Analysis*

All data are presented as means ± SD. Normality of the distribution of the data was examined with the Shapiro–Wilk's W test. A *t*-test was used to examine the differences among the medium-level and the low-level runners in the physiological parameters and race pace characteristics measured. A two-way analysis of variance with repeated measures in the second factor was used to examine the differences between the two groups in the mean velocity of each running split (0–10, 10–21.1, 21.1–30 and 30–42.195 km). Significant differences between means were located with the Newman–Keuls post hoc test. Pearson product moment correlations were used to determine the association between marathon time and the measured parameters. The statistical significance level was set for all tests at *p* < 0.05.
