How to dismantle modern stressors: does a short trip to simulated Paleolithic conditions in the wild reduce cortisol levels? [version 1; peer review: 1 not approved]

Background: Chronic stress has become a central problem of our modern society. It appears plausible that our Paleolithic ancestors were more exposed to acutely life-threatening stress stimuli than chronically enduring psychosocial stressors. The aim of this study was to test whether it is possible to reduce physiological and psychological stress parameters by returning to simulated Paleolithic lifestyle conditions. Methods: In total, 15 volunteers undertook a four-day hiking tour simulating hunter-gatherer lifestyle conditions in a National Park. Saliva samples were taken at 5, 6 and 7 pm prior to, during and after the intervention and evaluated for cortisol concentration. Perceived stress was measured by the Perceived Stress Questionnaire. Results: The study was completed by 11 of the subjects. Mean body weight and fat mass decreased significantly during the intervention by 2.3 and 2.2 kg, respectively. There was a marked increase in cortisol levels on the first day of the intervention compared to baseline with a gradual return to baseline levels on the subsequent days. Individual cortisol responses were heterogeneous; in threes subjects, cortisol concentrations generally decreased from 5 to 7 pm, while in four subjects the opposite trend was observed. Conclusions: During this four-day outdoor intervention under simulated Paleolithic conditions, significant changes occurred almost exclusively at the first day of the study. The increase in cortisol levels Open Peer Review


Introduction
Over the past 2.5 million years, humans, and the species preceding them, have continuously adapted to the corresponding environmental conditions 1 . For the longest period of this development, humans subsisted through a hunter-gatherer lifestyle 2 . However, environmental conditions changed ~12,000 years ago with the cultivation of plants and domestication of animals, and even more drastically ~150 years ago with the beginning of industrialization and 30 years ago with the advent of digitization 3,4 . Adaptation to stress signals is an integral part of human evolution 5 . The concept of the stress response is based on the idea that organisms activate protective mechanisms that allow them to survive unfamiliar and potentially dangerous situations 6 . Those who are resilient to internal and external danger signals increase their chance of survival, thereby also ensuring the survival of their genes 7 .
Chronic stress has become a central problem of our modern society, which is associated with numerous diseases, such as obesity, type 2 diabetes or depression 8 . It appears plausible that our Paleolithic ancestors were more exposed to acutely life-threatening stress stimuli than chronically enduring psychosocial stressors such as existential anxiety; indeed, many modern-day stressors including pressure to meet certain deadlines, social media stress etc. are a very recent phenomenon. Here, we report findings of a study that was designed to subject civilized humans to an imitation of the Paleolithic lifestyle for a short period of time. The research question was whether it is possible to reduce physiological and psychological stress parameters by returning to simulated Paleolithic conditions in a National Park over a four-day duration.

Design
The study was approved as a natural study by the ethics committee of the German Sports University Cologne (061-2018). All subjects signed a written informed consent concerning possible risks of participation before the start of the intervention. For a realistic simulation of Paleolithic conditions, the following basic requirements were defined: 1. During the intervention, subjects should remain exclusively in the wild.
2. In order to simulate hunting and gathering behavior, subjects should travel a distance of at least 20 km per day.
3. The first meal was not to be consumed before 12 am to simulate a period of hunting and gathering.
4. The diet should be based on the principles of a "Paleolithic diet" as defined by Cordain 9 , including lean meats, fish, fruits, vegetables, nuts, oils (coconut oil, olive oil) and sweet potatoes.

5.
The sleep-wake cycle should be adapted to the natural circadian rhythm according to sunrise and sunset.
The study lasted a total of five days and four nights from 25 th to 29 th July 2018. However, the last day was for return only, so the intervention spanned a total of four full days. The tour was led by Sascha Krüger (SK), who himself was a participant in our pilot studies published by Freese et al. 3,4 , and offers these tours commercially under the name "BACK-2BASIC-Tour". Our pilot studies took place in 2015 and 2016. The general study design was the same as in this follow-up study, but the research objective had focused on metabolic parameters.
At the beginning of the intervention, a 30-minute information session was held by the study director. Subjects were given extensive information about the exact procedure of the study as well as their tasks and were given the opportunity to receive answers to their open questions. During the entire intervention period, the study management team was available on site as scientific back-up. To analyze stress levels, saliva samples were taken before (t 0 ), during (t 1-4 ) and after (t 5 , t 7 , t 11 ) the intervention to determine cortisol levels.

Participants & recruitment
All participants in the commercially offered BACK-2BASIC tour were informed in advance about our study, and written informed consent was obtained. Exclusion criteria were any known diseases or the administration of medications that could affect cortisol levels. Participants completed a 30-minute on-site introductory seminar about the main principles and goals of the study.

Interventions
To simulate Paleolithic conditions, the following interventions were implemented during the study period: • Provision of "Paleolithic foods": At lunchtime participants received a food package consisting of two pieces of fruit and vegetables and 100 grams of nuts. Before sunset, we delivered a freshly prepared, Paleo-compliant meal to their place of residence.
• Spending 24 hours each day in an open arboreous environment • Cut off from technology and modern-style work stress (e.g., notifications from mobile phones, email, time pressure, etc.) • Exposure to the natural 24-hour temperature variability (only modest amount of clothing and sleeping bag was allowed) Hiking for several hours each day starting after sunrise The subjects were within the group at all times, supervised by the experienced tour guide SK. Simon Schnell (SS) was present each day at dinner to collect the saliva samples and hand out the new ones for the following day. Otherwise, there were no contacts during the interaction. Since subjects were constantly outside, all participants inevitably had to adhere to the study guidelines.

Measurements
Health Questionnaire and Anthropometric Data. Prior to the start of the intervention, each subject completed a health questionnaire that collected the following parameters: Personal, anthropometric and physical activity data, dietary behavior, health complaints and diseases. In addition to the requested anthropometric data, body weight, fat-free mass, skeletal muscle mass, body fat mass, and body mass index (BMI) were determined using a bioimpedance scale (Tanita model DC 430 MA P). At the end of the intervention, measurements on the bioimpedance scale were repeated. A copy of the questionnaire is provided as extended data 10 .
Saliva samples. Three days before, during, and seven days after the 4-day intervention (i.e., from t0-t11), three saliva samples were collected from each subject at 5, 6, and 7 pm, respectively. Saliva samples were collected in Salivettes Cortisol Code Blue (Sarstedt, Nümbrecht, Germany) and analyzed for cortisol concentrations in the laboratory of the Institute of Medical Psychology and Behavioral Immunobiology, Universtity Hospital Essen, using a commercially available enzyme-linked immunosorbent assay (Cortisol ELISA, IBL International, Hamburg, Germany) according to the manufacturer's instructions as previously described 11,12 . Intra-and inter-assay variance was 4.8% and 5.9% respectively, the detection limit was 0.005 μg/dL.

Psychological Stress Parameters.
Perceived stress was measured every evening at the time of saliva sampling by the German version of the Perceived Stress Questionnaire 13 . The PSQ contains 20 items and comprises four subscales (worries, tension, joy and demands). The questionnaire's instruction asked the participants to rate how often an item applied to their life within the last four weeks or last four days, respectively. The rating scale ranges from 1 ("almost") to 4 ("usually"). An overall score of perceived stress can be built by a summation of the four subscales. A pre-post intervention comparison of the psychological stress parameters (perceived stress and mood) was conducted using dependent t-tests.

Demographic profile
Out of 15 recruited volunteers, 11 completed the study (7 female and 4 male). Two were unable to attend for organizational reasons, while two others left during the intervention due to illness. The mean age of those who completed the study was 40.9 years (range 27-59 years). None of the participants was on medical treatment and all were considered mentally and physically healthy.

Paleolithic simulation conditions
Due to experiences of our studies from 2013-2014, we did not measure the average kcal intake per day and the macronutrient composition during the four nights and days in the wild. Average temperature was 33° Celsius at day one. On the other days, average temperatures ranged between 26-29° Celsius. The average daily hiking distance was 25.3 km.

Anthropometric and biochemical measurements
Four days of simulated Paleolithic conditions resulted in significant changes in subjects' body composition. Body weight dropped on average from 78.5 kg to 76.2 kg (-2.9%, p = 1.185 × 10 -13 ). Body fat mass was reduced from 20.8 kg to 18.6 kg (-2.3 kg, p = 0.0002), which corresponds to a reduction of the body fat percentage from 26.3% to 24.2% (-2.1%). All changes in body composition were highly significant (see Table 1 10 ).
Cortisol measurements from t 0 -t 11 Friedman's multi-factorial analysis of variance confirmed that there were differences in cortisol levels between the individual measurement days (p = 7.19 × 10 -16 ). As a result, post-hoc tests were performed taking the time of sampling into account. An initial analysis of average cortisol levels throughout the day (see Figure 1) revealed a marked increase in salivary cortisol at t 1 with a slight increase to t 4 . As it turned out, nearly all significant changes in cortisol levels were related to t 1 . Within 12 cases of significant changes, t 1 was involved ten times. Figure 1 shows cortisol levels over the individual measurement days, according to the time of sampling. The curve at 5 pm shows at t 1 a lower increase in cortisol compared to 6 pm and 7 pm (p=0.00146 and p=0.000706, respectively). Nevertheless, cortisol level at 5 pm displays a significant difference between t 1 and t 7 and between t 4 and t 7 .
The mean cortisol level at t 1 increased almost threefold from 5 to 6 pm (from 10.82 to 29.5 ng/ml). The cortisol value at 6 pm on the first day (t 1 ) differed significantly from all other 6 pm cortisol values, except on t 4 . Consistent with the general trend, the 6 pm cortisol curve shows a significant increase from t 0 to t 1 , followed by a rapid drop at t 2 . On the other hand, the slight increase in salivary cortisol at t 4 does not differ significantly from the measurements on the other days at 6 pm.  At 7 pm, the highest cortisol levels of the study were measured at t 1 (mean 34.83 ng/ml, maximum 170 ng/ml). Most of the significant changes at 7 pm are related to t 1 . In addition, 7 pm cortisol levels differ significantly from t 4 to t 7 , as in the measurements at 5 pm. Regardless of sample-taking time, there were no significant differences in the pre-post comparison. Neither at t 5 , t 7 nor t 11 significant changes from the initial value at t 0 could be found. The only significant changes from t 0 occurred at t 1 . Even during the intervention itself, there were hardly any significant changes in cortisol levels. Only between t 1 and t 3 (6 pm) and between t 1 and t 2 (7 pm) cortisol levels decreased significantly.
Changes in cortisol Measurements between 5 pm, 6 pm and 7 pm To investigate differences in cortisol values between 5-6 pm, 6-7 pm and 5-7 pm, a variance analysis was performed. The outcome was positive (p=7.19×10 -16 ). Post-hoc tests showed significant differences in cortisol levels over several days, depending on the time of the day, but not on all days. At t 0 , t 1 , t 2 , t 3 and t 11 , cortisol changed significantly in the course of time. In contrast, there were no significant differences in cortisol measured at the different times of the day at t 4 , t 5 and t 7 .
In two out of eight days, there was a significant change in cortisol levels between 5 and 6 pm (at t 1 and t 3 ). Between 6 pm and 7 pm cortisol levels did not vary significantly. Most of the changes were detected between 5 and 7 pm (at t 0 , t 1 , t 2 , t 3 and t 11 ), especially during the intervention period lasting from t 1 -t 4 .
As Figure 1 shows, cortisol levels decreased from 5 to 6 pm and from 6 to 7 pm on six out of eight days, except for t 1 and t 4 . Nevertheless, as shown in Figure 2, mean salivary cortisol levels showed a continuous increase from 5-6 pm (4.32 ng/ml to 5.9 ng/ml) and 6-7 pm (5.9 ng/ml to 6.33 ng/ml).

Cortisol measurements of individual subjects
Cortisol levels were also examined individually in all subjects. For this purpose, subjects were divided into "pro"responders and non-responders. The former were comprised of subjects whose cortisol levels generally decreased from 5 to 7 pm (on all days), whereas subjects whose cortisol levels generally increased from 5 to 7 pm were rated as non-responders. In total, three subjects were classified as pro-responders, four subjects as non-responders. Noteworthy, cortisol levels of pro-responders increased significantly at the beginning of the study (see Figure 3). At baseline (t 0 ), the average concentration of salivary cortisol was 8.27 ng/ml at 5 pm, 5.25 ng/ml at 6 pm and 4.01 ng/ml at 7 pm. Non-responders had already low cortisol levels at the beginning of the study (see Figure 4).
In a pre-post comparison, pro-responders achieved a significant reduction in cortisol levels. From t 0 to t 11 , average cortisol decreased from 8.27 to 2.18 ng/ml at 5 pm, from 5.25 to 1.37 ng/ml at 6 pm, and from 4.01 to 1.95 ng/ml at 7 pm. In contrast, cortisol levels of non-responders increased between t 0 and t 11 , albeit to a lesser extent than pro-responders decreased. Four subjects could not be classified as pro-or non-responders. These neither had a consistent increase nor a reduction of cortisol levels in the pre-post comparison.

Discussion
The aim of our study was to test how living in the wild for a short period of time affects markers of stress, which is an important research question given that nowadays many people spend significant amounts of time in highly artificial environments.
Daily hiking tours averaging 25.3 km per day exposed subjects to physical stress. Research has shown that medium to high intensity exercise (60-80% VO 2 max) leads to an increased cortisol output, whereas moderate exercise (40% VO 2 max) reduces cortisol 16 . Walking is generally classified as a low-intensity exercise. However, due to the long hiking distances and the uncommon high ambient temperatures, individual activity levels may have been strenuous for most of the subjects. Brenner et al. 17 revealed a cumulative effect of moderate-to high-intensity exercise and heat on cortisol output. In addition, they pointed out that repetitive physical stress at high temperature conditions causes elevated stress reactions, which last for longer periods after cessation of exercise. This outcome may explain the sharp and progressive increase of cortisol levels at t 1 .
Due to the high ambient temperature and the above-average intensity for most subjects on the first day of the intervention, it should be considered that the large increase in cortisol levels at t 1 may be due to synergistic effects of heat and exercise. In three of 11 subjects, cortisol levels at t1 showed no changes at all. Obviously, these subjects may be accustomed to prolonged stress in the heat. Health questionnaires confirmed this assumption: two of these subjects regularly participated in similar events, especially the tour guide who completed this intervention for the third time with other groups that year. One subject regularly performed endurance training. According to Martikainen et al. 18 endurance training results in a reduced activity of the hypothalamus-pituitary-adrenal (HPA) axis with correspondingly lower cortisol release. For all other subjects, who were not accustomed to the extensive physical workload in combination with heat and intermittent fasting conditions, the cumulative lifestyle changes would likely have been a strong stressor.

Conclusions
During this four-day outdoor intervention under simulated Paleolithic conditions, significant changes occurred almost exclusively at the first day of the study (t 1 ). The increase in cortisol levels at this point supports the hypothesis that non-responders respond by an increased release of cortisol as a result of adaptation to new environmental conditions. Although we could not find an average pre-post change in this small sample size, the individual analysis shows encouraging results. The distinction between pro-and contra-responders displays that subjects with high psychosocial stress could benefit from such an intervention. Future studies should recruit a larger number of subjects with more relevant inclusion criteria, such as subjects, who are permanently exposed to high levels of psychosocial stress.
As a limitation of this study, all analyses should be considered with caution, due to the small sample size and heterogeneity of the study participants.  number of analyses you want to correct for; some authors choose to correct for the number of analyses in one table, other authors correct for the number of analyses in the whole paper. In this paper, table 1 has four analyses, so the Bonferroni corrected significance value is p = 0.05/4 = 0.0125. Comparing weight before and after the intervention with an independent t-test results in p = 4.26 x 10 -7 . So, weight loss is significant in this study because this p-value is below the significance value of p = 0.0125.
I would suggest adding a measurement of variability to all figures, e.g. standard deviation, standard error of the mean or 95% confidence intervals.

5.
Whenever exact numbers are depicted in the text of the results section or the abstract, a measurement of variability, e.g. standard deviation, should be added. However, if the exact numbers are described in the table, you do not need to repeat those numbers in the text of the results section.

6.
It should be added to the figure legend of table 1 that 'mean ± standard deviation' is depicted in this table. In table 1, the standard deviation should be added for the difference.

7.
Because of the significant weight loss in this study, it would be valuable to report calorie intake.

8.
Several participants left the intervention early. Are there weight and cortisol measurements available for those participants? Did those participants differ from the other participants that completed the intervention?

9.
It is highly debated if salivary cortisol is a good biomarker for stress; for chronic psychological stress it may not be a biomarker at all. The authors should add a couple of sentences about the usefulness of their biomarker to the discussion section. An example of a paper that discusses this: Hellhammer DH et al: Salivary cortisol as a biomarker in stress research. Psychoneuroendocrinology (2009) 34, 163-171 1 . 10.