Before discussion of the present results, we will describe the meteorological characteristics of the three cities, Osaka, Poznan and Chiang Mai, and the definition of fasting RQ from the point of carbohydrate and fat metabolism.
Table 1 shows the monthly mean temperature and precipitation in the years of 1981 to 2010 (the average) and the year of the RQ examination. On average, we can categorize each two-month period according to the order of mean temperatures as follows: in Osaka (Japan) and Poznan (Poland); Jul to Aug (the hottest season, summer), Oct to Nov (autumn), Apr to May (spring) and Jan to Feb (the coldest season, winter). In Chiang Mai, they were categorized according to monthly mean temperatures and precipitation: Apr to May (hot season), Jul to Aug (rainy season), Oct to Nov (dry season), and Jan to Feb (dry and cool season). According to the World map of Koppen-Geiger climate classification , Osaka is located in a warm oceanic climate zone, Poznan is in the continental climate zone, and Chiang Mai is in a tropical monsoon climate zone.
The definition of the fasting RQ value is the RQ value in the postabsorptive state, where the postabsorptive state is typically represented by the state after an overnight fast before breakfast is consumed. In this phase, all of the meal has been absorbed from the intestinal tract . In this study, participants came to the experimental room from their residences by their usual means of transportation having performed their usual physical activities before the RQ examination, but having fasted for 12 hours or more. Their RQ values were therefore classed as fasting RQ values in everyday life. In a postabsorptive state, the concentrations of glucose and insulin are at their lowest and the concentration of non-esterified fatty acids is at its highest in a day . In this study, we neglected the contribution of protein oxidation to total RQ values by not measuring urea excretion because nutrients oxidized in well-nourished people are mostly carbohydrates and lipids . In addition, it is important to note that, in this postabsorptive state, glucose that enters the blood is almost exclusively from the liver (a proportion arises from glycogen breakdown and a proportion from gluconeogenesis) and non-esterified fatty acids are released by the action of hormone-sensitive lipase on the triacylglycerol stores in adipose tissue. Therefore, the nutritional composition of food ingested in the previous day or the day before (in a short-term) does not directly affect the fasting RQ values.
Seasonal variation in the fasting respiratory quotient values
There was a significant seasonal variation in the participants in Osaka, and the mean RQ value in Jul to Aug was significantly lower than that in Jan to Feb (the lowest mean value in the four seasons); those values were in the order of Jan to Feb > Oct to Nov > Apr to May > Jul to Aug. Table 5 shows the bivariate correlation of the fasting RQ and four items of food intake survey, indicating no correlation between them in Japanese participants. This lack of correlation between the RQ values and the items of food intake suggests that the nutritional composition of foods ingested had no effect on the fasting RQ values in the long-term in Japanese participants. In addition, there were no seasonal change in energy, fat or carbohydrate intake, as well as C/F ratio (see Table 3). These facts indicate that there are some factors other than food intake affecting the seasonal variation in the fasting RQ values.
Using the RQ values obtained in this study, we can calculate and estimate the ratio of fat oxidation to the total oxidation of carbohydrate and lipids by the following equation: (1.00-RQ)/(1.00-0.70). Application of this equation gives the ratios of carbohydrates metabolized to lipids metabolized in the four seasons in the participants in Osaka as follows: Jan to Feb (0.46: 0.54), Oct to Nov (0.44: 0.56), Apr to May (0.37: 0.63) and Jul to Aug (0.34: 0.66). These ratios indicate that relatively more lipids were metabolized in Jul to Aug (summer) and Apr to May (spring) compared to in Oct to Nov (autumn) and Jan to Feb (winter). These results agree with several reports about human seasonal variation in fasting serum glucose and triglyceride levels. For example, Behall et al.  observed that fasting serum glucose level was in the order of Dec to Feb > Sep to Nov > Mar to May > Jun to Aug in in young women living in California, USA and Gordon et al.  reported that there was a significant fall in triglyceride level from summer to winter in participants living in London, UK. This seasonality reported in the fasting serum glucose and lipid levels is consistent with the seasonal variation in the fasting RQ values observed in this study; in winter, the more fasting serum glucose there is, the more glucose is metabolized and the higher the obtained fasting RQ values are. In contrast, in summer, the more fasting triglyceride there is, the more fat is metabolized, the lower the obtained RQ values are. There are, however, few reports about seasonality in fasting serum glucose and plasma triglyceride levels of modern Japanese young female participants. In order to make it clear that our postulation described above is one of the possible mechanisms of the seasonal variation in the fasting RQ values observed in Japanese participants, we need further examination of seasonal variation in fasting serum glucose and triglyceride levels of the Japanese population, and seasonal change in longer-term control of gene expressions related to carbohydrate and lipid metabolism.
In the participants of Poznan, as described in the report of Tsumura et al. , there was significant seasonal variation in the efficiency of carbohydrate absorption in the intestine similar to that of Japanese participants; however, there was no significant seasonal variation in the fasting RQ values in this study. Our speculation, described in the Background, is that seasonal change in the efficiency of carbohydrate absorption in the intestine may have a relationship with the balance of carbohydrate and fat metabolism as observed in Japanese participants. Comparison of the fasting RQ value between Japanese and Polish participants shows that the mean RQ value of Polish participants in Jul to Aug was relatively higher than that of the Japanese participants. These higher RQ values in Jul to Aug might result in a non-significant seasonal variation in the fasting RQ values. One possible reason for the observed relatively higher RQ values in the summer period might be related to the significantly higher carbohydrate intake during Jul to Aug periods (see Table 3), on the assumption that this affects the carbohydrate and fat metabolism accompanied by other seasonal factors, such as changes in temperature and/or light intensity, which are peculiar to Polish living environments. Here, we can refer to studies by Plasqui et al.  and Oshiba , who reported that seasonal environmental changes affect total thyroxine or thyroid-stimulating hormone levels that influence human metabolic rate (probably the carbohydrate/fat metabolism). Of course we need to have more evidence concerning this assumption.
In Thai participants, there was no seasonal variation in the fasting RQ values. This could be explained by our previous finding that there was no significant seasonal variation in the AUCD of young female Thai university students .
Seasonal variation in the percent body fat and its relation to the seasonality in the fasting respiratory quotient values
In this study, we analyzed the percent body fat measured by an electric bioelectrical impedance analysis scale to provide an estimation of the relationship between the seasonality of fasting RQ values and that of the percent body fat in each population. The lack of a simple and international standard method for evaluation of exact body composition constituted a limitation to our study.
In Japanese participants, the seasonality of the percent body fat decreased in the order Jan to Feb > Oct to Nov > Apr to May > Jul to Aug. The percent body fat in the period of Jul to Aug was significantly lower than that in the other three periods. As described in the Background, Yamashita et al.  found that there was a significant seasonal change in the percent body fat among female university students living in metropolitan areas, where it is highest in winter and lowest in summer, which is consistent with the results obtained in this study. In addition, a similar seasonality of body fat percentage was reported by Mori et al. . They investigated seasonal change of fat mass in thirteen women of 20 to 30 years old using magnetic resonance imaging and reported that body weight, percent body fat, the subcutaneous fat mass and subcutaneous fat thickness of the abdomen were highest in January and lowest in July. These previous reports and our present result suggest that modern young Japanese females still have some regulatory mechanism of fat accumulation according to seasonal changes in their circumstances. This mode of seasonal changing in the percent body fat, decreasing in spring to summer and increasing in autumn to winter, was similar to that of the fasting RQ values in Japanese participants.
This similarity in seasonal changes between the fasting RQ values and the percent body fat observed in Japanese participants may represent a possible physiological mechanism. We postulate that Japanese participants may metabolize more fat to lose fat storage in the hot season, but metabolize more carbohydrate to save fat for storage in the colder season under an almost constant energy intake and expenditure, as described in the Background section. This constancy was shown in Table 3, which demonstrated that there was no significant seasonal variation in the energy intake of Japanese participants, and by our previous study, which showed there was no significant seasonal variation in the normal daily physical activity of Japanese university female students, as measured by a small accelerometer . Our hypothesis might be supported by a report concerning the seasonal variation in adipose tissue lipoprotein lipase, which provides free fatty acids for storage in adipocytes; where it was higher in winter than in summer as observed in participants in Colorado, USA . In order to examine this postulation, we calculated within-participant correlation coefficients between the fasting RQ values and the percent body fat, on the assumption that the correlation will be significant if our postulation is one of the possible regulatory mechanisms. However, Table 5 shows that there was no significant correlation within participants between the fasting RQ values (as the independent variable) and the percent body fat (dependent variable) in Japanese participants. This result indicates that seasonal variation in the balance of carbohydrate and fat metabolism has no significant effect on the seasonality in the percent body fat.
In Polish participants, there was a significant seasonal variation in the percent body fat; the order was that of Oct to Nov > Jan to Feb > Apr to May > Jul to Aug. This mode of changing was very similar to that of Japanese participants, where the percent body fat was the lowest in Jul to Aug and relatively higher in Oct to Nov and Jan to Feb. To examine the correlation between the seasonality in the fasting RQ values and the percent body fat in Polish participants, we calculated correlation coefficient values within participants and found that there was no significant correlation between them (Table 5). This result, together with that of Japanese participants, contradicts our postulation. In this correlation analyses, however, the correlation coefficient values of Japanese and Polish participants were nearly close to 0.200 (which is generally accepted as the least value suggesting very weak correlation). The borderline probability in both participant groups may allow us to speculate that there was a big seasonal effect on the balance of carbohydrate and fat metabolism, and so on fat accumulation, in our ancestors long ago, but the effect has become so small that it no longer has any influence on our percentage of body fat, particularly because our ancestors spent their daily life in a much more food-scarce and less artificial environment than the modern human population. Modern human beings, especially those living in cosmopolitan areas, are surrounded by artificial environments, such as the constant temperature provided by air-conditioning and foods available at any time. This speculative theory is supported by other examples of the loss of seasonality in our metabolic physiology: Nakamura et al.  reported that seasonal variations in basal metabolic rate were getting small, especially in indoor sedentary workers; and Maeda et al.  reported that there was no seasonal variation in the basal metabolic rate in university students. These authors considered that the improvement of air-conditioning influences the artificial microclimate and might reduce the seasonal variation in the basal metabolic rate of the Japanese population.
One more possible factor that affected the present correlation analysis is the participants’ degree of obesity (or leanness). There is a possibility that the serum non-esterified fatty acid level relates to the amount of fat storage in the adipose tissue, that is, the degree of obesity. We should have analyzed within-participant correlation coefficients between the fasting RQ values and the percent body fat, but we could not analyze the correlation statistically under the stratification of the degree of obesity due to the small number of the participants participated in this study.
In Thai participants, there was also a significant seasonal change in the percent body fat; the order was that of Oct to Nov > Jan to Feb > Jul to Aug > Apr to May. It is noteworthy that there was a common feature of the mode of seasonal change in the percent body fat in the three different populations, where that percent body fat was relatively lower in the hot season than in the colder seasons. (In Chiang Mai, the hotter season is Apr to May while the colder season is Jan to Feb, see Table 1.)
Regarding body weight, there was a significant seasonal change in the Polish participants’ body weight, which was in the same order as that in the percent body fat. There were, in contrast, no significant seasonal changes in those of Japanese and Thai participants, although there were significant seasonal changes in the percent body fat. This could be explained by the fact that we measured the participants’ body weight including their clothes (without their outer wear). The body weight without clothes of Japanese and Thai participants were lower than those of Polish participants, so that the even a small seasonal change in the weight of the clothes worn would affect body weight. We are now re-investigating the seasonal change in body composition, including the percent body fat, body weight and muscular mass, of Japanese and Thai participants wearing the same clothes for every examination throughout a year.
Seasonal variation in macronutrient intake
As summarized in Table 3, the results of macronutrient intakes of the two participant groups reflected characteristics of Japanese and Polish dietary habits. Polish participants take in more energy, protein and fat than Japanese participants do, while Japanese participants take in more carbohydrates than Polish participants do. Detailed discussion about the present food intake survey in Japanese and Polish participants will be published elsewhere, using food intake records of 100 Japanese and 111 Polish participants. The points discussed here are the seasonal changes in fat and carbohydrate intake, in addition to those in the C/F ratio. As shown in Table 3, there was no significant seasonal variation in the C/F ratio in Japanese participants. Owaki et al.  reported similar results, with no significant seasonal variation in total energy intake or in the ratio of energy intake from protein, fats and carbohydrates (similar to C/F ratio). In Japan, the disappearance of a seasonal variation in food intake has accelerated in recent years due to rapid development of the food transportation system and food production control including intensive culture by house and frozen storage.
In Polish participants, the C/F ratio in Apr to May was significantly lower than that in Jul to Aug, which reflected a significant decrease in the percentage of total energy from carbohydrate intake, and a non-significant increase in energy from fat intake Apr to May. This increase in fat intake in spring (and in winter) was also reported by Przysiężna and Banachowicz  with female students of the Wroclaw University of Economics. Their percentage of fat intake in the total energy intake was 32 % in winter, 32 % and 30 % in autumn. This increase in fat intake may be related to the effect of high-fat diet on thermal acclimation proposed by Yoshimura et al.  as a metabolic adaptation to cold temperatures in winter. Therefore, we should take into account the effect of ingestion of food rich in fat on the increase in the percent body fat in the colder seasons in Polish participants. These facts may be reflected in the relatively higher correlation coefficient within participants (although not significant) between the fat intake (as independent variable) and the percent body fat (dependent variable) in Table 5.