Next Article in Journal
Young People’s Perspectives on Climate Change in Urban Brazil
Previous Article in Journal
McDonald’s McLean Deluxe and Planetary Health: A Cautionary Tale at the Intersection of Alternative Meats and Ultra-Processed Marketing
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Challenges
Volume 16
Issue 3
10.3390/challe16030034
challenges-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Moderate Impact of Increasing Temperatures on Food Intake in Human Populations

by
Per M. Jensen
* and
Marten Sørensen
Department of Plant and Environmental Sciences, University of Copenhagen, Thorvaldsensvej 40, 1871 Frederiksberg, Denmark
*
Author to whom correspondence should be addressed.
Challenges 2025, 16(3), 34; https://doi.org/10.3390/challe16030034
Submission received: 11 June 2025 / Revised: 17 July 2025 / Accepted: 18 July 2025 / Published: 21 July 2025
(This article belongs to the Section Human Health and Well-Being)
Browse Figures
Versions Notes

Abstract

Increasing temperatures associated with climate change will lead to (periodic) temperature-induced reductions in food intake in human and other mammal populations. Human adults, however, are both tolerant and resilient to periodic nutritional deficits, and the associated health effects should be limited. Intermittent nutritional deficits may also cause growth restriction in developing foetuses and young children, which potentially affects their food intake in later life. Therefore, temperature-induced hypophagia can be hypothesised to manifest as later compensatory responses with multiple concomitant (or extended) lags of varying temporal dimensions. We examined the relationship between calorie intake and ambient outdoor temperatures for a time series covering past decades (FAO data for 1961–2013) in 80 countries to determine if humans alter their food intake in response to elevated temperatures. We included eleven different temporal “windows of exposure” of varying lag. These windows considered current and recent exposure, just as lagged effects allowed for a consideration of past effects on mothers, their children, and childhood exposure. It was hypothesised that one of these could provide a basis for predicting future changes in human calorie intake in response to climate change. Our analyses showed no apparent association with temperatures in ten of the eleven hypotheses/models. The remaining hypothesis suggests that current calorie intake is linked to decadal mean temperatures with a lag of approximately three decades, pointing to an impact on mothers and their (developing) children. The impact of an increase in mean temperature varies with temperature amplitudes, and negative impacts are only found in countries with low temperature amplitudes (warmer countries), albeit the impact on calorie intake caused by a 2–3 °C change in temperatures or temperature amplitudes is generally modest. However, in considering calorie intake, we only address quantities of food (with unspecified quality), which insufficiently reflect the full range of nutritional challenges associated with increasing temperatures. Understanding climate-driven changes in human food intake requires global interdisciplinary collaboration across public health, environmental science, and policy.

1. Introduction

In recent decades, researchers have made considerable efforts to assess the challenges arising from climate change [1,2]. Most studies focus on the indirect effects associated with concurrent environmental change [3,4], while a smaller but still substantial number address the direct effects on humans. Direct effects are typically linked to the acute impacts imposed on human physiology by extreme temperatures, manifesting, for example, as increased mortality among elderly individuals who are less able to cope with heat stress for extended periods [5,6]. More subtle effects are linked to negative impacts on human reproduction [7,8], as well as on pregnancies and foetal growth [7,9], which may alter development trajectories and health in adulthood. Other subtle effects, linked to mental health [2], might affect human behaviour and social interactions. However, the scale and mode of action are rarely clear, and we cannot assess the true importance of these effects.
Similar uncertainties shroud the possible impact of temperature-induced hypophagia and its associated effects on human health. There is no doubt that elevated temperatures can affect human food intake [10], just as they can in all other medium-sized mammals [11,12,13,14]. Our understanding of the relationship between ambient (annual mean) temperature and food intake builds on classical studies [15,16], which emphasise that animals change in shape and size to accommodate the energetic cost of living in cooler environments, as well as the detrimental effects of intermittent or chronic heat stress [17,18]. The calorie intake will reflect these environmental challenges, and we expect higher calorie intake among people living in the Arctic who suffer the highest heat loss (Figure 1A) [10,19,20].
The experimental data suggest that food intake declines as the annual mean ambient temperatures increase from 0 to 10 °C. Food intake is stable between 15 and 20 °C and slightly lowered between 10 and 15 °C, if or when food intake serves as an alternative method of heat production [21,22,23]. Food intake further diminishes with the suspension of other heat-generating activities when temperatures increase above approx. 25 °C. While we know that these principles apply, we must also note that the sensitivity to changes in temperatures is affected by physical size and activity patterns. Larger and physically active individuals tend to have both a lower sensitivity to cooler temperatures and a higher sensitivity to elevated temperatures. That is, larger individuals have a lower thermoneutral zone [24,25] (Figure 1A). Other anatomical and behavioural traits, such as varying inclination to engage in additional food intake for heat production and differences in ability to store and dissipate heat [26,27], could also have significant effects. Thus, we should question whether the association noted in short-term experimental studies (Figure 1A) also applies to human populations distributed worldwide. Perhaps we should even insist that local adaptation should be our point of departure and, hence, that past adaptation led to differences in size and shape of humans [28,29], meaning that individuals living in cooler climates eat more than individuals in warmer climates simply because they are larger [30]. So, what appears to be a fair alignment between expected (Figure 1A) and reported food intake (food availability) from FAO (Figure 1B) could be misleading.
Several studies on seasonal variation in food intake [31] clearly show that the seasonal variation in food intake for various food groups can be substantial (>10%, Figure 2A). When the observed data for the seasonal variation are viewed in the context of the national calorie intake records from FAO for the relevant populations [32], populations with low food intake have the highest seasonal variation. Hence, people living under substantial seasonal temperature amplitudes exhibit a slight seasonal variation in calorie intake, which depends on physical activity [33], but maintain a steady, high daily consumption (Figure 2B,C). The same can be noted for urban populations in the USA, with exceptionally high calorie intake [34,35]. Unsurprisingly, the slight perpetual seasonal oscillations in food intake are associated with limited seasonal weight cycling [36]. For populations in warmer climates, the opposite is true (Figure 2B,C [33]), which aligns with the idea that it is extended periods with elevated temperatures that are associated with a lowered calorie intake (Figure 2B,C). It is, however, not readily accepted that seasonal variation in food intake (seasonal hypophagia) can lower the total annual food intake. Adults exposed to food deprivation will reduce their physical activity but become leaner, while fat stores are depleted [37,38]. During weight regain, they temporarily overcompensate and achieve a higher weight and BMI than before starvation [38]. However, they will gradually reach the same weight, BMI, and food intake as they originally had, indicating that periods of food restriction do not affect mean annual calorie intake unless the calorie deprivation occurs in one year, and the weight regains in the next. Our inclination to consider the remarks on the seasonal variation in food intake is further diminished by the fact that a part of the noted seasonal variation (Figure 2A), at least in some cases, could arise from actual food shortages [33].
However, the tolerance and resilience to periodic temperature-induced hypophagia do not extend from mothers to developing foetuses because foetal nutrition varies with placental blood flow during acclimation [7,9]. Hence, recurrent or permanently diminished blood flow and nutritional delivery could lead to intrauterine undernutrition, which manifests as reduced birth weights [39,40,41]. Intrauterine growth restriction may alter foetal development and promote an array of postnatal changes in food intake and metabolism [42,43]. It could be hypothesised that intrauterine growth restriction and weight cycling in childhood [44,45,46,47] may promote altered absorption and post-absorption processes, enabling the achievement of development targets [48] on a reduced calorie intake [49]. With this, our expectation of a thermal effect on calorie intake becomes a question of how temperatures contribute to differences in birth weights and foetal development [50].
Considering the importance of birth weights and, hence, the “developmental origins of health and disease” hypothesis [51,52,53] seems prudent, as the hypothesis increasingly has been used to explain a variety of metabolic and behavioural traits [54,55]. Applying it to address how entire populations modulate food intake due to temperature changes would seem beyond its scope. Moreover, it does not seem likely that shifts in birth weights could be detected in calorie intake for an entire population because each annual cohort of newborns typically constitutes approx. 1% of the population (assuming ten births per one thousand inhabitants). The only time when these individuals could make a detectable mark in the total population’s intake would be during their early teenage years, when weight gains [56] and physical activity [57] would be substantial for children that grow tall (high birthweight) [58] and less for those who would not (low birth weight). We will, without hesitation, acknowledge that high birth weight neonates become heavier adults with higher food intake, which influences their food intake in youth [30,59]. But we still find it challenging to accept that temperatures, through their impact on birth weight, affect food intake. The reason is that a host of other factors, especially socioeconomic development, influence the year-by-year variation in birth weight [60]. However, if we ignore these critical comments, we will note that temperatures undoubtedly influence birth weights, and thermal cues, mediated by maternal effects, are likely to be a key contributor to global differences in birth weights [61]. These effects are moderated by intergenerational inertia [62,63], which protracts phenotypic responses due to biological constraints. Little is known of the temporal integration of thermal cues and how such contributions to reductions in foetal growth might alter future food intake. Albeit if, for example, mothers were influenced by thermal cues accumulated from birth until the time they gave birth (approx. 30 y, [64]), then the association between temperatures and calorie intake in their offspring would present itself with a lag of several decades.
Our initial consideration was that any impact of temperatures on human food intake per inhabitant would be undetectable and more likely would be associated with other causes, such as changes in age distribution and physical size [30,65,66], changes in physical work [30], infectious diseases [67,68], and societal/socioeconomic perturbations [69,70]. However, examining various time series on birth weight and calorie intake did not support such critical considerations (Figure 3A–C). Hence, we hypothesised that temperatures experienced by mothers could affect future food intakes for entire populations through changes in birth weight and associated changes in adult anthropometrics, behaviour, and metabolism [65]. As the variation in calorie intake appeared to be associated with birth weights 10–15 ys earlier (Figure 3), we expected that the average lag could be as much as 45 y.
This paper examines the relationship between annual calorie intake per inhabitant per day (FAO data [32]) and ambient outdoor temperatures for time series covering past decades (1961–2013) in 80 countries worldwide. It contributes to the current literature by examining the relationship between calorie intake and temperature for highly aggregated data in a global context, which enables a critical assessment of studies that have examined local or regional short-term associations. The analysis was designed to question the relationship between mean temperatures and human food intake (Figure 1A). Concurrently, we acknowledged that the socioeconomic developments and the long-term compensatory effects on the maternal offspring could be of greater importance [63,71,72]. We included eleven models that defined specific “lifetime windows of exposure” of varying length and lag. It was hypothesised that at least one of these “windows” and temporal lags would identify an association between past temperatures and current calorie intake, which would allow us to determine whether temperature-dependent calorie intake follows the relationship given in Figure 1A or Figure 2C, and whether there is a substantial lag (decades) for the impact of temperatures on human food intake.

2. Materials and Methods

2.1. Analytical Approach

Our analysis departs from the well-known relationship between food intake and mean temperatures (Figure 1A). The relationship is challenging in statistical terms because it is difficult to define suitable statistical models for a 3rd–4th order polynomial function. It is thus quite fortunate that the mean temperatures in all current nations with geographically even distributed populations have high mean temperatures, encompassing temperatures from approx. 10 to 35 °C (Figure 1B). Geographically adjusted mean temperatures are lower in Canada and Russia. However, these should/must be dismissed/ignored because most of the population resides in the warmer parts of the countries. In doing so, we will find that the relationship between daily calorie intake per capita can be assumed to align with a 2nd-order polynomial relationship (Figure 1C). If this is true, then we must expect to find that the change in calorie intake per degree is a 1st-order relationship (Figure 1C), which is a perfectly suitable hypothesis to investigate by simple generalised linear models (e.g., y = x2 ↔ y′ = 2x; see examples in Barber [73] and Jensen et al. [8]). To pursue this line of inquiry, we first extracted the change in calorie intake per degree change in temperatures (in extraction models) for many countries, and second, we analysed the dependency between reaction norms and temperatures (in assessment models). In preparation, we first considered whether putative contributors were likely to change in concordance with temperatures (and bias the reaction norms) or whether they could be accepted as a state variable that, independently of temperature changes, affected calorie intake.

2.2. Separation of Variables in State and Non-State Variables

We could start by extracting the correlation coefficient between the mean temperatures and the calorie intake for further analysis. However, we may also use slightly more complex models and incorporate variables that we suspect could interfere with the outcome. Children have a lower intake than adults [30], and differences in age structure will result in different estimates for temperature-dependent calorie intake for growing and stable populations. We therefore added the median age of the populations to our models. Human height and weight also affect calorie intake [65], but the change in phenotype could be considered a direct response to temperature-induced calorie intake. By not adjusting for height and weight, we retained the option to detect the long-term effects of temperature. However, we considered both minimum and maximum temperatures to study the effects of temperature amplitudes. Other effects associated with infectious diseases [67,68], emotional stress [74], and economic disruptions, among others, were relevant to consider, as these also would deliver prudent adjustment to the reaction norms. However, we chose to accept these as state variables because they either change very slowly or, in equal terms, depend on the suite of changes that follow gradual demographic and socioeconomic development. Socioeconomic conditions were in the context of the five decades covered in the data extraction (1961–2013) and considered a state variable, in line with variables like latitude, altitude [75,76], ethnicity/culture/lineage [77], etc.

2.3. Extraction of Reaction Norms and AIC Values from Time Series

The reaction norms were extracted using minimum and maximum temperatures, which were adjusted for the changes in the median age in the populations. The base of extraction was given as: Calorie-intake = a × Min-temp + b × Max-temp + c × median age + err.
Calorie intake was reported as mean calorie intake per day−1 and per capita−1 in the year. Min-temp was the monthly minimum temperature within the year or an equivalent running mean over several years. Max-temp was the monthly maximum temperature within the year or an equivalent running mean temperature over several years, and median ages were the 50th quartile for entire populations in the given year [78]. We defined eleven data aggregations/windows of exposure (Table 1), which permitted an assessment of systematically changing the running means and corresponding lags. Each of the eleven models delivered estimates for the effects of temperatures (a and b), which correspond to the effects of a one–degree increase in the minimum and maximum temperatures, respectively (RN-Min and RN-Max). The impact of one degree in both represents the reaction norm to a two-degree increase in mean temperatures (RN-Mean). The Akaike information criteria (AIC) value was noted for each model to assess the variation in model quality.

2.4. The Applicability and Quality of Assessment Models

Our ability to explain the differences among the national reaction norms is critical for accepting the appropriateness of the given extraction model (window width and lag, Table 1, Figure 1C). However, the reaction norms will inevitably increase and potentially become inflated when the number of years included in the running means spanning the “lifetime window” increases. The reason is that the calorie intake (the nominator) remains at the same scale of variation. In contrast, a steadily diminishing temperature variation (the denominator) explains the variation in calorie intake. This imbalance leads to enhanced stratification of the reaction norms, but it also means that estimated effects become “inflated” when the “window” is widened beyond measure. It may be difficult or impossible to ensure that the window width is sufficiently “inclusive” and applies to all populations/nations without generating inflated reaction norms. Splitting exceedingly long “lifetime” periods into multiple windows (the period is covered by several shorter windows/running means) should be a suitable method of addressing these challenges because both the summed and individual reaction norms from each window can be assessed in the same manner as we would evaluate the reaction norms from a single window. Our ability to account for the variation in reaction norms also depends on the length of the time series included. Shorter time series will undoubtedly be more sensitive to socioeconomic perturbations, leading to inaccurate reaction norms when they coincide with temperature changes. It is, therefore, important that time series are kept at their maximum length. A window that covers 30 years, with a lag of 30 years, requires 60 years of data prior to the year of observation (of the independent variable). With a record for temperatures starting in 1901, the maximum permissible temporal displacement (combined window width and lag) is 60 years, provided that the first year of observation in 1961 must be included.
While the rigid, one-size-fits-all approach may lead to inaccurate estimates, the target for our analysis remains well-defined due to its simplicity. We hypothesised that reaction norms diverge for populations residing in either cooler or warmer climates, indicating that people are differentially affected by increased temperatures (Figure 1C). A similar effect of maximum temperatures (a) and minimum temperatures (b, a = b) is required to accept that the association between food intake and mean temperature is a 2nd-order polynomial function. An inverse association (a = −b, or −a = b) would suggest that temperature amplitudes modulate the mean reaction norm.
It is also noteworthy that model bias can be assessed by statistical analyses that examine the systematic variation in the Akaike information criteria (AIC) among extraction models. AIC values will be high (indicating inferior quality) in countries where factors other than temperatures and median age lead to changes in calorie intake and where the data quality is poor. This will affect all models for a given country because none of the extraction models includes the relevant information. Moreover, an analysis of the relative AIC values (AICn/mean AIC mean(1–11)) can identify whether countries with specific attributes have a poorer goodness of fit. Such analyses can, for example, detect whether temperatures have a lower ability to explain the variation in food intake in high-income countries where people’s exposure to ambient weather might be lessened. Ideally, there should be no significant associations between relative AIC values and any state variable. We would dismiss the outcome if the relative AIC values depended on temperatures.

2.5. Data Sets for Extraction Models

Information on calorie intake per day−1 per capita−1 was retrieved from FAO national food balance sheets [32,79,80] covering the years from 1961. FAO’s methodology was modified in 2014, which might have changed the assessments. Due to this uncertainty, we included records from 1961 to 2013 (53 years). Countries that did not include information over the given years were excluded. The per capita intake used corresponds to a three-year, rather than an annual, average, which evens out the effect of errors in the annual food stock data used [79], albeit the data quality is moderate and associated with several assumptions and uncertainties [81,82]. It is, however, essential to note that any country-specific fixed bias is eliminated in the extraction models.
The temperature data originated from the World Bank Group Climate Knowledge Portal [83]. The appropriate minimum (minimum mean monthly temperature within the year(s)) and maximum (maximum mean monthly temperature within the year(s)) temperatures were extracted for the period 1901 to 2013, and running means for each of the eleven hypotheses were calculated. The temperature records represent geographically adjusted monthly mean temperature for entire countries, which will have limited accuracy in larger countries with geographically unevenly distributed populations. Canada was excluded due to the geographically uneven population distribution. Data for median ages originated from Our World in Data [78].
We adopted a national time-series approach alphabetically and intended to include half of the nations with adequate data (1961–2013; 80 of 159) to determine whether this was sufficient to address the hypotheses. We found the eighty records to be sufficient, which means that our extraction models encompass countries from Afghanistan to Mongolia. The total number of observations was 4240 (80 × 53 y). These formed the basis for 880 extraction models (11 models × 80 countries), which delivered separate estimates for the effects of minimum and maximum temperatures. Reaction norms (RN-mean, RN-Min, and Rn-Max) were calculated and subjected to analysis. When the models combined two windows, the reaction norms were calculated as the average effect from both.

2.6. Data Sets for Assessment Models

The assessment models included the following state-variables: mean temperatures (minimum and maximum temperatures over the period 1961 to 2013), socioeconomic conditions (high income, upper middle income, lower middle income, and low-income countries [84]), mean height (mean of male and female height, cm) in the period 1961 to 2013 [85], the proportion of the population living below 500 m elevation above sea level [86], and the mean calorie intake per day−1 per person−1 for the period 1961 to 2013 [32]. The given variables were used to assess systematic association for reaction norms and model quality (AIC values).

2.7. Validation by Standard Cross-Sectional Analysis

As the assessment models are unlikely to deliver accurate estimates for the impact of changing temperatures, we close our presentation by constructing a cross-sectional analysis based on our simplified time-series analysis described above. It serves as validation, as we would reject the outcome of our time-series analyses if the inferred models did not have a cross-sectional analogue. We ensured that the cross-sectional analysis covered less than half the period and twice the number of countries, such that the cross-sectional model incorporated data for a period where the state variables would show minor variation and encompassed countries excluded in the time-series analysis. The cross-sectional model included the interactions with temperatures noted in the previous analyses. The model was reduced to include only the significant contributors (p < 0.05), and the predicted global calorie intake was compared with the outcome of the simplified time series analysis. FAO provided information on national calorie intake for 174 countries for the years 1990–1992, 1995–1997, 2000–2002, and 2006–2008 [87], and we used the mean for the four periods for analysis. The number of observations was reduced to 162 due to missing data and the exclusion of larger countries (Canada, China, and the USA). All analyses were performed using SAS 9.4 (SAS Institute, Cary, NC, USA).

3. Results

Our analysis of the mean reaction norms (RN-Mean) generated by the 11 extraction models (Table 1) identified significant associations with temperatures (p < 0.05) in three models (Model 5, 3030, and D30), indicating that the remaining models may be considered irrelevant to our hypotheses. The models gave variable levels of explanation (R2) ranging from 0.04 to 0.37 (Table 2), which in most cases were generated by associations with other variables other than temperatures. In comparing the R2s for models with and without maximum and minimum temperatures, we found modest increments in the R2s for adding temperatures, except in Models 3030 and D30 (Figure 4A), suggesting that only these two models were relevant for consideration. The analysis of the individual 30 and 3030 windows within Model D30 suggested that only the latter was associated with temperatures (Table 2). The RN-Means for Model 3030 were significantly associated with long-term minimum and maximum temperatures (Table 3). A negative association was observed for minimum temperatures and a positive association for maximum temperatures.
There was an apparent effect of socioeconomic conditions and height on RN-Mean (Table 3). Moreover, a sizable effect of both minimum and maximum temperatures on the RN-Mean led to a varying effect across temperature amplitudes (Table 3, Figure 4B). The estimated effect of temperatures on RN-Mean remained much the same in models with and without other state variables (Table 3 and Table 4), which suggested that the impact of socioeconomic conditions and height resulted from confounding between the two, i.e., these state variables explain variation in RN-Mean, which was unrelated to the contribution from temperatures. Consistently, the analysis of relative AIC values for Model 3030 only showed a decreasing model quality for nations with a high proportion of people residing at high altitudes (Table 3). Subsequent analyses of the RN-Max and RN-Min within Model 3030 revealed that the association with temperatures remained significant for RN-Max, whereas no associations were found for RN-Min. The difference between RN-Max and RN-Min remained when these were analysed for an association with temperatures without supporting state variables (Table 4).
Challenges 16 00034 g004
Figure 4. Results from the analysis of calorie intake. (A) Output from the global variation in reaction norm analysis for eleven models, each representing a specific hypothesis on the association between temperatures and calorie intake (Table 1). The level of explanation (R2) is shown for eleven assessment models, with and without temperatures. (B) The variation in RN-mean for eighty countries vs. mean temperature amplitudes (see the effect of temperatures in Table 3). (C) Predicted calorie intake day−1 person−1 predicted by temperatures alone (Table 5) for a low-income population with a median age of 25 y living at sea level (0 m asl). The dotted lines show the full range of predictions, while the complete lines show combinations of mean temperatures and temperature amplitudes encompassed by the current climates. Out of bounds: The models’ assumption does not permit predictions of effects for annual mean temperatures less than approx. 10 °C.
Figure 4. Results from the analysis of calorie intake. (A) Output from the global variation in reaction norm analysis for eleven models, each representing a specific hypothesis on the association between temperatures and calorie intake (Table 1). The level of explanation (R2) is shown for eleven assessment models, with and without temperatures. (B) The variation in RN-mean for eighty countries vs. mean temperature amplitudes (see the effect of temperatures in Table 3). (C) Predicted calorie intake day−1 person−1 predicted by temperatures alone (Table 5) for a low-income population with a median age of 25 y living at sea level (0 m asl). The dotted lines show the full range of predictions, while the complete lines show combinations of mean temperatures and temperature amplitudes encompassed by the current climates. Out of bounds: The models’ assumption does not permit predictions of effects for annual mean temperatures less than approx. 10 °C.
Challenges 16 00034 g004
The standard cross-sectional analysis identified a significant interaction between mean, minimum, and maximum temperatures, as indicated in the analysis of reaction norms (Table 3 and Table 5). No temperature-dependent interactions were noted for height, altitude, or socioeconomic conditions. However, each contributed independently. The independent estimates for the minimum and maximum temperatures were similar, and the sum matched the negative contribution for the mean temperatures. Therefore, the summed effect of a change in mean temperatures was slight, leaving temperature amplitudes to generate substantial systematic differences worldwide.
The projected change in calorie intake per degree of change in mean temperatures was modest to negligible (Figure 4C). However, the model accounted for a sizable part of the global variation in calorie intake, which was attributed to the significant differences in temperatures worldwide. As a change in temperature amplitudes was associated with changes in calorie intake, the cross-sectional model corroborated the association between temperatures and calorie intake, as shown in Model 3030.

4. Discussion

In our suite of models, we assessed the possible impact of temperatures on the current calorie intake in human populations. An array of models was included because the impact of temperatures is likely to have varying effects on physiological and phenotypic traits, depending on the age and time of exposure. The analyses’ outcome seems meagre, as they, with just one exception, fail to detect any relevant contributions of temperatures to global differences in calorie intake (Table 2, Figure 2A). The approach has several limitations, which might account for this. First and most importantly, we committed ourselves to the assumption that calorie intake would align clearly with mean temperatures and designed the analyses accordingly (Figure 1C). Proceeding with this notion would have required both maximum and minimum temperatures to have a similar effect in the assessment model, which was not the case (Table 3). The outcome might have been clearer if the extraction models had included temperature amplitudes. However, this approach would not have allowed us to reject the notion that there is a second-order relationship between mean temperatures and calorie intake. Second, the significant effects of both socioeconomic conditions and height on RN-Mean (Table 3) underline that there are systematic and complex effects of the ‘standard of living’ (i.e., income and height). The outcome suggests that wealthy populations tend to lower their calorie intake as they age, which may be linked to declining physical activity in affluent countries. Thus, the approach might benefit from incorporating other temporal dynamics, like financial crises and technological advances, into the extraction models. Such improvements limit the variations in reaction norms among countries with similar climates but could add further challenges.
However, in criticising the statistical approach, we would soberly acknowledge that we did not fail to acknowledge the substantial effects of socioeconomic conditions (Table 3 and Table 5 [70]). Economic development primarily affects calorie intake in human populations by modulating population size and structure and permitting increases in human growth [88,89]. The noted RN-Mean was frequently associated with economic conditions (achieved in 2017 [84]) because most countries experienced significant economic development concurrently with increasing temperatures (Figure 4A). These positive associations are unrelated to climate change, and neither do they seem to modulate the effect of ambient temperatures (Table 5). Further differences in the outcomes between the temporal and cross-sectional analyses (Table 3 and Table 5) likely arise from the addition of numerous countries with higher temperature amplitudes (East European countries) in the standard cross-sectional analysis (Table 5), which unhinge the associations (confounding) noted for the 80 reaction norms (Table 3).
The association between temperatures and calorie intake, as presented in our cross-sectional analysis (Table 5), has high credibility because we accounted for the temporal cause-and-effect relationship in clear terms. The RN-mean is modulated by maximum temperatures (Table 3 and Table 4), while the variation in minimum temperatures is less clear or contingent on maximum temperatures. The outcome suggests that minimum temperatures contribute less accurately to the model prediction (Table 3 and Table 4), as expected when the compensatory action (minimum temperature) only delivers an effect in proportion to previously imposed temperature-induced hypophagia (maximum temperatures). The global pattern depicted in Figure 4C corresponds well with the association between mean temperatures and global birth weights shown by Roberts [39]. In acknowledging the association with birth weights, we suspect that the predictions are suboptimal because our current analyses do not include daily temperature amplitudes, which contribute to the global variation in birth weights [61,90]. This shortcoming can also be recognised from the reduced model quality (higher AIC-values) for high-altitude populations (Table 3), as they experience higher daily temperature amplitudes than populations residing at sea level [90,91].
Our “surviving” hypothesis states that current calorie intake can be predicted from temperatures decades before the present (Model 3030, Table 2 and Table 3), which include an average lag of 45 years for any given year. We find no unambiguous evidence for an impact of concurrent or recent temperatures on current calorie intake. However, there are indications of such effects in Model 5. If this partially supported model was accepted as credible, it would also imply that, in recent years, high maximum temperatures (causing hypophagia) have increased calorie uptake years later [38]. The response type in Model 5 concurs with our ‘surviving hypothesis’ by indicating that past periods with temperature-induced nutrient limitation led to a later extended or lasting compensatory action. We find that temperature-induced reduced food intake (hypophagia) leads to a later increment in calorie uptake, i.e., it has the same consequences as undernutrition induced by other causes [38,42,45]. With this, we could propose that the given patterns emerge from a negative impact of elevated temperatures on developing children [40,41] and seasonal weight cycling in early childhood [45,46], and that the effect on food intake and metabolism emerges in adulthood [42,43]. The wider association with temperature amplitudes arises from seasonal effects, which at higher latitudes, turn autumn weight gains [33,36] into spring growth [92,93]. The seasonality of growth may be different and more pronounced in warmer countries.
Temperatures account for a fair part of the global range in calorie intake (approx. 1500 to 3700 calories; Figure 1B). Thus, an increase in mean temperatures from 10 to 35 °C accounts for differences of approx. 1700 to 2500 calories (Figure 4C), due to a temperature-dependent linkage with body size and changes in metabolism. Biological factors, such as local adaptations and lineage, will also contribute [26,27,94]. Assuming the association remains stable under a change in climate, we can expect an increase or decrease in calorie intake to depend on temperature amplitudes (Figure 4). In general, an increase in mean temperatures of one degree will be associated with a reduction of 32 calories per day−1 per person−1 (800 calories/25 °C), i.e., less than the calorie content in a standard 33 cc soft drink. We would consider this impact modest compared to the impact of socioeconomic development (Figure 2C, Table 5), even if we accepted that the impact can be up to 50 calories per degree (depending on the temperature amplitude; Table 5, Figure 4C) and that mean temperatures may increase by 2–3 °C. Thus, the indirect effects of climatic change, e.g., the negative impact on agricultural yields [3,4,95,96,97], must be of far greater importance. Such concerns are compounded by past and current socioeconomic development, which, due to improved diet quality, have produced larger human body sizes with higher calorie requirements [88,89], and who, at least in theory, will be more susceptible to the negative impact of elevated temperatures. Thus, this study contributes to the growing body of evidence on the health implications of climate change within the planetary health paradigm [98], which recognises that human health and well-being are inextricably linked to the health of Earth’s natural systems. It underscores that human health is not a reflection of the current, but rather a product of past, planetary health.

Strengths and Limitations

The strengths of our assessment lie in the amount of data included and the number of hypotheses that are considered, as these mitigate concerns that spurious associations were identified. The outcome will have a global impact and apply to various regions, as general principles are universally applicable. At the national level and for communities, these may be of little relevance because other factors promote or limit food security. It is also important to note that calorie intake is just one aspect of human food intake, and that food preferences and the associated modulation of food quality may be temperature dependent. Analyses of calorie intake cannot address such concerns, as they primarily focus on energetic challenges.

5. Conclusions

We find (i) a clear but modest effect of raising mean temperatures on current calorie intake in human populations. Some populations will see an increase in calorie intake, others a reduction, and some will not change their calorie intake (ii). These patterns arise because the effect of mean temperature is modulated by temperature amplitudes and/or maximum temperatures, which vary systematically with mean temperatures globally. (iii) The impact is recognised as an average effect affecting the global population, over several decades, manifesting with a lag of approx. three decades. This lag suggests an impact on mothers, their developing children, and infants. Hence, the outcome aligns with the “developmental origins of health and disease hypothesis”, which in recent decades has shaped the understanding of current health challenges in many developed countries. To firmly establish that temperatures systematically and differentially affect calorie intake through the modulation of foetal development will, however, be challenging, due to ethical limitations. Nevertheless, future studies should address these potential impacts and also investigate broader nutritional aspects, enabling a more comprehensive assessment of the potential adverse effects on human nutrition, growth, and development. Ensuring sufficiently contrasting outcomes for individuals and populations may require studies with substantial geographical coverage.

Author Contributions

P.M.J.: Conceptualisation, Data curation, Formal Analysis, Methodology, Visualisation, Writing—original draft. M.S.: Visualisation, Writing—review and editing. All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Data Availability Statement

No new data were produced in this study.

Acknowledgments

This study would not have been possible without free access to data records from FAO, WHO, UN, Gapminder, Our World in Data, and The World Bank Climate Knowledge Portal.

Conflicts of Interest

The authors declare no conflicts of interest.

References

  1. IPCC. Climate Change 2023: Synthesis Report, Summary for Policymakers. Contribution of Working Groups I, II and III to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change; Lee, H., Romero, J., Eds.; IPCC: Geneva, Switzerland, 2023. [Google Scholar]
  2. Rocque, R.J.; Beaudoin, C.; Ndjaboue, R.; Cameron, L.; Poirier-Bergeron, L.; Poulin-Rheault, R.A.; Fallon, C.; Tricco, A.C.; Witteman, H.O. Health effects of climate change: An overview of systematic reviews. BMJ Open 2021, 11, e046333. [Google Scholar] [CrossRef]
  3. Iizumi, T.; Ramankutty, N. Changes in yield variability of major crops for 1981–2010 explained by climate change. Environ. Res. Lett. 2016, 11, 034003. [Google Scholar] [CrossRef]
  4. Katona, P.; Katona-Apte, J. The interaction between nutrition and infection. Clin. Infect. Dis. 2008, 46, 1582–1588. [Google Scholar] [CrossRef] [PubMed]
  5. Ebi, K.L.; Capon, A.; Berry, P.; Broderick, C.; de Dear, R.; Havenith, G.; Honda, Y.; Kovats, S.; Ma, W.; Malik, A.; et al. Hot weather and heat extremes: Health risks. Lancet 2021, 398, 698–708. [Google Scholar] [CrossRef] [PubMed]
  6. Diniz, F.R.; Gonçalves, F.L.T.; Sheridan, S. Heat wave and elderly mortality: Historical analysis and future projection for metropolitan region of São Paulo, Brazil. Atmosphere 2020, 11, 933. [Google Scholar] [CrossRef]
  7. Hansen, P.J. Effects of heat stress on mammalian reproduction. Philos. Trans. R. Soc. B: Biol. Sci. B 2009, 364, 3341–3350. [Google Scholar] [CrossRef]
  8. Jensen, P.M.; Sørensen, M.; Weiner, J. Human total fertility rate affected by ambient temperatures in both the present and previous generations. Int. J. Biometeorol. 2021, 65, 1837–1848. [Google Scholar] [CrossRef]
  9. Choudhari, R.H. Multidimensional Impact of Climate Change on Human Reproduction and Fertility: A Medical Perspective on Changing Dynamics. In Research Anthology on Environmental and Societal Impacts of Climate Change: 1672–1709; IGI Global: Hershey, PA, USA, 2022. [Google Scholar] [CrossRef]
  10. Johnson, R.E.; Kark, R.M. Environment and food intake in man. Science 1947, 105, 378–379. [Google Scholar] [CrossRef]
  11. Fuquay, J.W. Heat stress as it affects animal production. J. Anim. Sci. 1981, 52, 164–174. [Google Scholar] [CrossRef]
  12. Nienaber, J.A.; Hahn, G.L.; Yen, J.T. Thermal environment effects on growing-finishing swine Part I—Growth, feed intake and heat production. Trans. ASAE 1987, 30, 1772–1775. [Google Scholar] [CrossRef]
  13. Renaudeau, D.; Gourdine, J.L.; St-Pierre, N.R. A meta-analysis of the effects of high ambient temperature on growth performance of growing-finishing pigs. J. Anim. Sci. 2011, 89, 2220–2230. [Google Scholar] [CrossRef] [PubMed]
  14. Thornton, P.; Nelson, G.; Mayberry, D.; Herrero, M. Increases in extreme heat stress in domesticated livestock species during the twenty-first century. Glob. Change Biol. 2021, 27, 5762–5772. [Google Scholar] [CrossRef] [PubMed]
  15. Bergmann, C. Uber die Verhaltniesse der Warmeokonomie der Thiere zu ihrer Grosse. Gott. Stud. 1847, 1, 595–708. [Google Scholar]
  16. Allen, J.A. The influence of physical conditions in the genesis of species. Radic. Rev. 1877, 1, 108–140. [Google Scholar]
  17. Bindon, J.R.; Baker, P.T. Bergmann’s rule and the thrifty genotype. Am. J. Phys. Anthropol. 1997, 104, 201–210. [Google Scholar] [CrossRef]
  18. Foster, F.; Collard, M. A reassessment of Bergmann’s rule in modern humans. PLoS ONE 2013, 8, e72269. [Google Scholar] [CrossRef]
  19. Leonard, W.R.; Sorensen, M.V.; Galloway, V.A.; Spencer, G.J.; Mosher, M.J.; Osipova, L.; Spitsyn, V.A. Climatic influences on basal metabolic rates among circumpolar populations. Am. J. Hum. Biol. 2002, 14, 609–620. [Google Scholar] [CrossRef]
  20. Yurkevicius, B.R.; Alba, B.K.; Seeley, A.D.; Castellani, J.W. Human cold habituation: Physiology, timeline, and modifiers. Temperature 2022, 9, 122–157. [Google Scholar] [CrossRef]
  21. Brobeck, J.R. Food intake as a mechanism of temperature regulation. Obes. Res. 1997, 5, 641–645. [Google Scholar] [CrossRef]
  22. McCue, M.D. Specific dynamic action: A century of investigation. Comp. Biochem. Physiol. Part A Mol. Integr. Physiol. 2006, 144, 381–394. [Google Scholar] [CrossRef]
  23. Secor, S.M. Specific dynamic action: A review of the postprandial metabolic response. J. Comp. Physiol. B 2009, 179, 1–56. [Google Scholar] [CrossRef] [PubMed]
  24. Kingma, B.; Frijns, A.; van Marken Lichtenbelt, W. The thermoneutral zone: Implications for metabolic studies. Front. Biosci.-Elite 2012, 4, 1975–1985. [Google Scholar] [CrossRef] [PubMed]
  25. Pallubinsky, H.; Schellen, L.; van Marken Lichtenbelt, W.D. Exploring the human thermoneutral zone—A dynamic approach. J. Therm. Biol. 2019, 79, 199–208. [Google Scholar] [CrossRef] [PubMed]
  26. Taylor, N.A. Ethnic differences in thermoregulation: Genotypic versus phenotypic heat adaptation. J. Therm. Biol. 2006, 31, 90–104. [Google Scholar] [CrossRef]
  27. Hanna, J.M.; Brown, D.E. Human heat tolerance: An anthropological perspective. Annu. Rev. Anthropol. 1983, 12, 259–284. [Google Scholar] [CrossRef]
  28. Roberts, D.F. Body weight, race and climate. Am. J. Phys. Anthropol. 1953, 11, 533–558. [Google Scholar] [CrossRef]
  29. Roberts, D.F. Climate and Human Variability, 2nd ed.; Cummings: Menlo Park, CA, USA, 1978. [Google Scholar]
  30. Tontisirin, K.; de Haen, H. Human Energy Requirements; Report of a Joint FAO/WHO/UNU Expert Consultation 17–24; FAO: Rome, Italy, 2001; Available online: https://www.fao.org/3/y5686e/y5686e.pdf (accessed on 13 March 2024).
  31. Stelmach-Mardas, M.; Kleiser, C.; Uzhova, I.; Peñalvo, J.L.; La Torre, G.; Palys, W.; Boeing, H. Seasonality of food groups and total energy intake: A systematic review and metaanalysis. Eur. J. Clin. Nutr. 2016, 70, 700–708. [Google Scholar] [CrossRef]
  32. FAO. Food and Agriculture Organization of the United Nations; 2021 Caloric Intake Per Day Per Capita. 2021. Available online: http://www.fao.org/faostat/en/#data/FBSH (accessed on 23 April 2021).
  33. Westerterp-Plantenga, M.S. Effects of extreme environments on food intake in human subjects. Proc. Nutr. Soc. 1999, 58, 791–798. [Google Scholar] [CrossRef]
  34. Ma, Y.; Olendzki, B.C.; Li, W.; Hafner, A.R.; Chiriboga, D.; Hebert, J.R.; Campbell, M.; Sarnie, M.; Ockene, I.S. Seasonal variation in food intake, physical activity, and body weight in a predominantly overweight population. Eur. J. Clin. Nutr. 2006, 60, 519–528. [Google Scholar] [CrossRef]
  35. Bernstein, S.; Zambell, K.; Amar, M.J.; Arango, C.; Kelley, R.C.; Miszewski, S.G.; Tryon, S.; Courville, A.B. Dietary intake patterns are consistent across seasons in a cohort of healthy adults in a metropolitan population. J. Acad. Nutr. Diet. 2016, 116, 38–45. [Google Scholar] [CrossRef]
  36. Westerterp, K.R. Seasonal variation in body weight: An experimental case study. J. Therm. Biol. 2001, 26, 525–527. [Google Scholar] [CrossRef]
  37. Keys, A.; Henschel, A.; Mickelsen, O.; Taylor, H.L. The Biology of Human Starvation; University of Minnesota Press: Minneapolis, MN, USA, 1950; pp. 1–2. [Google Scholar]
  38. Dulloo, A.G. Physiology of weight regain: Lessons from the classic Minnesota Starvation Experiment on human body composition regulation. Obes. Rev. 2021, 22, e13189. [Google Scholar] [CrossRef]
  39. Roberts, D.F. Race, genetics and growth. J. Biosoc. Sci. Suppl. 1969, 1, 43–67. [Google Scholar] [CrossRef]
  40. Wells, J.C.; Cole, T.J. Birth weight and environmental heat load: A between-population analysis. Am. J. Phys. Anthropol. 2002, 119, 276–282. [Google Scholar] [CrossRef] [PubMed]
  41. Jensen, P.M.; Sørensen, M. Differences in human birth weight and corollary attributes as a result of temperature regime. Ann. Hum. Biol. 2013, 40, 385–395. [Google Scholar] [CrossRef] [PubMed]
  42. Hales, C.N.; Barker, D.J. The thrifty phenotype hypothesis: Type 2 diabetes. Br. Med. Bull. 2001, 60, 5–20. [Google Scholar] [CrossRef] [PubMed]
  43. Grzęda, E.; Matuszewska, J.; Ziarniak, K.; Gertig-Kolasa, A.; Krzyśko-Pieczka, I.; Skowrońska, B.; Sliwowska, J.H. Animal Foetal Models of Obesity and Diabetes–From Laboratory to Clinical Settings. Front. Endocrinol. 2022, 13, 63. [Google Scholar] [CrossRef]
  44. Prentice, A.M.; Jebb, S.A.; Goldberg, G.R.; Coward, W.A.; Murgatroyd, P.R.; Poppitt, S.D.; Cole, T.J. Effects of weight cycling on body composition. Am. J. Clin. Nutr. 1992, 56, 209S–216S. [Google Scholar] [CrossRef]
  45. Yang, Z.; Huffman, S.L. Nutrition in pregnancy and early childhood and associations with obesity in developing countries. Matern. Child Nutr. 2013, 9, 105–119. [Google Scholar] [CrossRef]
  46. Kroke, A.; Liese, A.D.; Schulz, M.; Bergmann, M.M.; Klipstein-Grobusch, K.; Hoffmann, K.; Boeing, H. Recent weight changes and weight cycling as predictors of subsequent two year weight change in a middle-aged cohort. Int. J. Obes. 2002, 26, 403–409. [Google Scholar] [CrossRef]
  47. Zou, H.; Yin, P.; Liu, L.; Duan, W.; Li, P.; Yang, Y.; Li, P.; Yang, Y.; Li, W.; Zong, Q.; et al. Association between weight cycling and risk of developing diabetes in adults: A systematic review and meta-analysis. J. Diabetes Investig. 2021, 12, 625–632. [Google Scholar] [CrossRef] [PubMed]
  48. Tanner, J.M. Growth as a target-seeking function: Catch-up and catch-down growth in man. In Human Growth: A Comprehensive Treatise Volume 1 Developmental Biology Prenatal Growth; Springer: Boston, MA, USA, 1986; pp. 167–179. [Google Scholar] [CrossRef]
  49. van Wesenbeeck, C.F.; Keyzer, M.A.; Nubé, M. Estimation of undernutrition and mean calorie intake in Africa: Methodology, findings and implications. Int. J. Health Geogr. 2009, 8, 1–18. [Google Scholar] [CrossRef] [PubMed]
  50. Kramer, M.S. Determinants of low birth weight: Methodological assessment and metaanalysis. Bull. World Health Organ. 1987, 65, 663–737. [Google Scholar] [PubMed]
  51. Buklijas, T.; Al-Gailani, S. A fetus in the world: Physiology, epidemiology, and the making of fetal origins of adult disease. Hist. Philos. Life Sci. 2023, 45, 44. [Google Scholar] [CrossRef]
  52. Hollstein, T.; Basolo, A.; Ando, T.; Votruba, S.B.; Walter, M.; Krakoff, J.; Piaggi, P. Recharacterizing the metabolic state of energy balance in thrifty and spendthrift phenotypes. J. Clin. Endocrinol. Metab. 2020, 105, 1375–1392. [Google Scholar] [CrossRef]
  53. Wells, J.C. Natural selection and human adiposity: Crafty genotype, thrifty phenotype. Philos. Trans. R. Soc. B 2023, 378, 20220224. [Google Scholar] [CrossRef]
  54. Del Giudice, M. Early stress and human behavioral development: Emerging evolutionary perspectives. J. Dev. Orig. Health Dis. 2014, 5, 270–280. [Google Scholar] [CrossRef]
  55. Wells, J.C.; Cole, T.J.; Cortina-Borja, M.; Sear, R.; Leon, D.A.; Marphatia, A.A.; Murray, J.; Wehrmeister, F.C.; Oliveira, P.D.; Goncalves, H.; et al. Low maternal capital predicts life history trade-offs in daughters: Why adverse outcomes cluster in individuals. Front. Public Health 2019, 7, 206. [Google Scholar] [CrossRef]
  56. Tanner, J.M. Foetus into Man: Physical Growth from Conception to Maturity; Harvard University Press: Cambridge, MA, USA, 1990. [Google Scholar]
  57. Santos, J.L.; Martinez, J.A.; Martinez-Gonzalez, M.A.; Gibney, M.J.; Irala, J.; Kearney, J.; Varo-Cenarruzabeitia, J.J. Prevalence of Physical Activity During Leisure Time in the European Union. 2001. Available online: https://hdl.handle.net/10171/17570 (accessed on 22 March 2024).
  58. Andersen, L.G.; Ängquist, L.; Gamborg, M.; Byberg, L.; Bengtsson, C.; Canoy, D. NordNet Study Group. Birth weight in relation to leisure time physical activity in adolescence and adulthood: Meta-analysis of results from 13 Nordic cohorts. PLoS ONE 2009, 4, e8192. [Google Scholar] [CrossRef]
  59. Wakimoto, P.; Block, G. Dietary intake, dietary patterns, and changes with age: An epidemiological perspective. J. Gerontol. Ser. A Biol. Sci. Med. Sci. 2001, 56 (Suppl. S2), 65–80. [Google Scholar] [CrossRef]
  60. Kogan, M.D. Social causes of low birth weight. J. R. Soc. Med. 1995, 88, 611–615. [Google Scholar] [CrossRef] [PubMed]
  61. Jensen, P.M.; Sørensen, M. In Search of Environmental Factors Associated with Global Differences in Birth Weight and BMI. Am. J. Hum. Biol. 2025, 37, e70038. [Google Scholar] [CrossRef] [PubMed]
  62. Kuzawa, C.W.; Eisenberg, D.T. Intergenerational Predictors of Birth Weight in the Philippines: Correlations with Mother’s and Father’s Birth Weight and Test of Maternal Constraint. PLoS ONE 2012, 7, e40905. [Google Scholar] [CrossRef] [PubMed]
  63. Kuzawa, C.W.; Fried, R.L. Intergenerational memories of past nutritional deprivation: The phenotypic inertia model. In The Arc of Life: Evolution and Health Across the Life Course; Springer: Berlin/Heidelberg, Germany, 2017; pp. 7–20. [Google Scholar] [CrossRef]
  64. OECD. Age of Mothers at Childbirth and Age-Specific Fertility (SF2.3) 2023. Available online: https://www.oecd.org/els/soc/SF_2_3_Age_mothers_childbirth.pdf (accessed on 7 April 2023).
  65. WHO. World Health Organization. Human Energy Requirements: Report of a Joint FAO/WHO/UNU Expert Consultation: Rome, 17-24 October 2001; Food & Agriculture Org.: Rome, Italy, 2004; Volume 1.
  66. Perkins, J.M.; Subramanian, S.V.; Davey Smith, G.; Özaltin, E. Adult height, nutrition, and population health. Nutr. Rev. 2016, 74, 149–165. [Google Scholar] [CrossRef]
  67. Langhans, W. Bacterial products and the control of ingestive behavior: Clinical implications. Nutrition 1996, 12, 303–315. [Google Scholar] [CrossRef]
  68. Brown, W.E.; Bradford, B.J. Invited review: Mechanisms of hypophagia during disease. J. Dairy Sci. 2021, 104, 9418–9436. [Google Scholar] [CrossRef]
  69. Rosen, S.L.; Shapouri, S.; Global Economic Crisis Threatens Food Security in Lower Income Countries. Amber Waves: The Economics of Food, Farming, Natural Resources, and Rural America. 2009. Available online: https://ageconsearch.umn.edu/record/122562/files/06GlobalEconomic.pdf (accessed on 11 April 2024).
  70. Waibel, H.; Hohfeld, L. Poverty and nutrition: A case study of rural households in Thailand and Viet Nam. In The Asian ‘Poverty Miracle; Edward Elgar Publishing: Cheltenham, UK, 2016; pp. 215–250. [Google Scholar] [CrossRef]
  71. Emanuel, I.; Filakti, H.; Alberman, E.; Evans, S.J. Intergenerational studies of human birthweight from the 1958 birth cohort. 1. Evidence for a multigenerational effect. BJOG Int. J. Obstet. Gynaecol. 1992, 99, 67–74. [Google Scholar] [CrossRef]
  72. Lummaa, V.; Clutton-Brock, T. Early development, survival and reproduction in humans. Trends Ecol. Evol. 2002, 17, 141–147. [Google Scholar] [CrossRef]
  73. Barber, N. On the relationship between fertility and geographic latitude: A cross-national study. Cross-Cult. Res. 2002, 36, 3–15. [Google Scholar] [CrossRef]
  74. Tamashiro, K.L.; Hegeman, M.A.; Sakai, R.R. Chronic social stress in a changing dietary environment. Physiol. Behav. 2006, 89, 536–542. [Google Scholar] [CrossRef]
  75. Singh, D.S.N.; Sridharan, K.; Selvamurthy, W. Human nutrition at high altitude. Nutr. Bull. 1999, 24, 195–202. [Google Scholar] [CrossRef]
  76. Morel, O.E.; Aubert, R.; Richalet, J.P.; Chapelot, J.P. Simulated high altitude selectively decreases protein intake and lean mass gain in rats. Physiol. Behav. 2005, 86, 145–153. [Google Scholar] [CrossRef]
  77. Atkin, D. The caloric costs of culture: Evidence from Indian migrants. Am. Econ. Rev. 2016, 106, 1144–1181. [Google Scholar] [CrossRef]
  78. Gapminder. Median Age in Years. Available online: https://www.gapminder.org/data/ (accessed on 23 April 2021).
  79. FAO. Food and Agriculture Organization of the United Nations. An Introduction to the Basic Concepts of Food Security. Italy, Rome. 2008. Available online: https://www.fao.org/3/al936e/al936e00.pdf (accessed on 12 February 2023).
  80. FAO. Food and Agriculture Organization of the United Nations. Food Balance Sheets—A Handbook. Rome Italy. 2001. Available online: https://www.fao.org/documents/card/en/c/7d49abaa-eccf-5412-878a-8c190c46748a (accessed on 12 February 2023).
  81. Cialfa, E.; Turrini, A.; Lintas, C. A National Food Survey: Food Balance Sheets and Other Methodologies: A Critical Overview. In Monitoring Dietary Intakes; ILSI Monographs; Macdonald, I., Ed.; Springer: London, UK, 1991. [Google Scholar] [CrossRef]
  82. Thar, C.M.; Jackson, R.; Swinburn, B.; Mhurchu, C.N. A review of the uses and reliability of food balance sheets in health research. Nutr. Rev. 2020, 78, 989–1000. [Google Scholar] [CrossRef] [PubMed]
  83. TWBC. The World Bank Climate Knowledge Portal. Historical Data Covering 1901 to 2016. 2020. Available online: https://climateknowledgeportal.worldbank.org/download-data (accessed on 1 June 2020).
  84. Gapminder. Geographic Territories Frequently Used by Gapminder. Includes Geographical Regions and National Income Level as Given by World Bank Income Group in 2017. 2018. Available online: https://www.gapminder.org/data/ (accessed on 19 June 2018).
  85. OWD. Our World in Data (OWD). Height by Year of Birth 1896 to 1996. 2019. Available online: https://ourworldindata.org/human-height (accessed on 3 February 2019).
  86. Tremblay, J.C.; Ainslie, P.N. Global and country-level estimates of human population at high altitude. Proc. Natl. Acad. Sci. USA 2021, 118, e2102463118. [Google Scholar] [CrossRef] [PubMed]
  87. FAO. Food and Agriculture Organization of the United Nations. FAO Statistics Division. 2020. Available online: http://www.fao.org/faostat/foodsecurity/Files/FoodConsumptionNutrients.xls (accessed on 11 September 2020).
  88. Depenbusch, L.; Klasen, S. The effect of bigger human bodies on the future global calorie requirements. PLoS ONE 2019, 14, e0223188. [Google Scholar] [CrossRef]
  89. Grasgruber, P.; Sebera, M.; Hrazdíra, E.; Cacek, J.; Kalina, T. Major correlates of male height: A study of 105 countries. Econ. Hum. Biol. 2016, 21, 172–195. [Google Scholar] [CrossRef]
  90. Jensen, P.M.; Sørensen, M. Spatiotemporal variations in human birth weight are associated with multiple thermal indices. Atmosphere 2025, 16, 569. [Google Scholar] [CrossRef]
  91. Jang, Y.S.; Shen, S.F.; Juang, J.Y.; Huang, C.Y.; Lo, M.H. Discontinuity of diurnal temperature range along elevated regions. Geophys. Res. Lett. 2022, 49, e2021GL097551. [Google Scholar] [CrossRef]
  92. Marshall, W.A. Evaluation of growth rate in height over periods of less than one year. Arch. Dis. Child. 1971, 46, 414–420. [Google Scholar] [CrossRef]
  93. Kościński, K.; Krenz-Niedbała, M.; Kozłowska-Rajewicz, A. Month-of-birth effect on height and weight in Polish rural children. Am. J. Hum. Biol. 2004, 16, 31–42. [Google Scholar] [CrossRef]
  94. Pomeroy, E.; Stock, J.T.; Wells, J.C.K. Population history and ecology, in addition to climate, influence human stature and body proportions. Sci. Rep. 2021, 11, 274. [Google Scholar] [CrossRef] [PubMed]
  95. Macdiarmid, J.I.; Whybrow, S. Nutrition from a climate change perspective. Proc. Nutr. Soc. 2019, 78, 380–387. [Google Scholar] [CrossRef] [PubMed]
  96. Ray, D.K.; Gerber, J.S.; MacDonald, G.K.; West, P.C. Climate variation explains a third of global crop yield variability. Nat. Commun. 2015, 6, 5989. [Google Scholar] [CrossRef] [PubMed]
  97. Frieler, K.; Schauberger, B.; Arneth, A.; Balkovič, J.; Chryssanthacopoulos, J.; Deryng, D.; Elliott, J.; Folberth, C.; Khabarov, N.; Müller, C.; et al. Understanding the weather signal in national crop-yield variability. Earth’s Future 2017, 5, 605–616. [Google Scholar] [CrossRef]
  98. Myers, S.S. Planetary health: Protecting human health on a rapidly changing planet. Lancet 2017, 390, 2860–2868. [Google Scholar] [CrossRef]
Challenges 16 00034 g001
Figure 1. Calorie intake per inhabitant per day. (A) The general relationship between mean temperatures and food/calorie intake for humans and other medium-sized mammals. F-TNZ: the functional thermoneutral zone, which varies with body size and standard clothing. (B) Scatterplot for calorie intake day−1 person−1 (FAO data) vs. mean temperatures in 162 countries. (C) Graphical illustration of the analytical principles of a 2nd-order function of temperatures, where it follows that the change in calorie intake per degree will be a 1st-order function (y = x2 ↔ y′ = 2x). Analysis of individual time series (arrows) compares to subjecting a narrow temperature interval to analysis in our extraction models. Black and grey lines refer to two populations of different physical sizes. Out of bounds: the models’ assumptions do not permit predictions of effects for mean annual temperatures less than approx. 10 °C.
Figure 1. Calorie intake per inhabitant per day. (A) The general relationship between mean temperatures and food/calorie intake for humans and other medium-sized mammals. F-TNZ: the functional thermoneutral zone, which varies with body size and standard clothing. (B) Scatterplot for calorie intake day−1 person−1 (FAO data) vs. mean temperatures in 162 countries. (C) Graphical illustration of the analytical principles of a 2nd-order function of temperatures, where it follows that the change in calorie intake per degree will be a 1st-order function (y = x2 ↔ y′ = 2x). Analysis of individual time series (arrows) compares to subjecting a narrow temperature interval to analysis in our extraction models. Black and grey lines refer to two populations of different physical sizes. Out of bounds: the models’ assumptions do not permit predictions of effects for mean annual temperatures less than approx. 10 °C.
Challenges 16 00034 g001
Challenges 16 00034 g002
Figure 2. Seasonal variation in calorie intake. (A) The coefficient of variation (×100) in food intake for specific food groups noted for two to four seasons in 21 studies compiled by Stelmach-Mardas et al. [31] plotted against the calorie intake person−1 day−1 records from FAO in the year of publication. (B) Logarithmic transformed coefficient of variation in food intake plotted against the seasonal temperature amplitude in the given cities (highest mean monthly temperature minus the minimum mean monthly temperature, temperature data from The World Meteorological Association). (C) Total national calorie intake [21] and partial variation in calorie intake as provided by studies mentioned in Stelmach-Mardas et al. [31], plotted against seasonal temperature amplitudes in the relevant cities.
Figure 2. Seasonal variation in calorie intake. (A) The coefficient of variation (×100) in food intake for specific food groups noted for two to four seasons in 21 studies compiled by Stelmach-Mardas et al. [31] plotted against the calorie intake person−1 day−1 records from FAO in the year of publication. (B) Logarithmic transformed coefficient of variation in food intake plotted against the seasonal temperature amplitude in the given cities (highest mean monthly temperature minus the minimum mean monthly temperature, temperature data from The World Meteorological Association). (C) Total national calorie intake [21] and partial variation in calorie intake as provided by studies mentioned in Stelmach-Mardas et al. [31], plotted against seasonal temperature amplitudes in the relevant cities.
Challenges 16 00034 g002
Challenges 16 00034 g003
Figure 3. Birth weight and calorie intake. FAO data [21] paired with records of birth weights in the same countries 10 to 15 years earlier. (A) The United Kingdom, (B) Japan, and (C) Tanzania. Birth weight records were obtained from multiple sources.
Figure 3. Birth weight and calorie intake. FAO data [21] paired with records of birth weights in the same countries 10 to 15 years earlier. (A) The United Kingdom, (B) Japan, and (C) Tanzania. Birth weight records were obtained from multiple sources.
Challenges 16 00034 g003
Table 1. Overview of the models and suggested interpretation of the temporal reference.
Table 1. Overview of the models and suggested interpretation of the temporal reference.
Model NameCalculated AsUnderstood AsFor a 35-Year-Old
Person, the Years Refer to.
1Minimum and maximum temperatures in the given yearAn impact of temperatures in the current yearCurrent year
5Average minimum and maximum temperatures for the past 5 years A general impact of temperature given by the mean over past years. Recent years
10Average minimum and maximum temperatures for the past 10 yearsA general impact of temperature given by the mean over the past decade Last decade
20Average minimum and maximum temperatures for the past 20 yearsA general impact of temperature given by the mean over past decadesLast decades
30Average minimum and maximum temperatures for the past 30 yearsA general impact of temperature from childhood to adulthoodCurrent generation
1010Average minimum and maximum temperatures for the past 10 years 10 years previouslyAn impact of temperatures over a decade, a decade ago (in recent past)Young adulthood
2020Average minimum and maximum temperatures for the past 20 years 20 years previouslyAn impact of temperatures over decades, decades ago (in the past)Young adulthood and some years prior to birth
3030Average minimum and maximum temperatures for the past 30 years 30 years previouslyA general impact of temperature on childhood and on the parents. Effects on parents and some years after birth.
D10Model 10 and 1010 combinedA general impact of temperature given by the mean over past yearsLast decades
D20Model 20 and 2020 combinedA general impact of temperature given by the mean over past decadesCurrent generation and some years prior to birth
D30Model 30 and 3030 combinedAn impact of temperature on both current and past generations.A lifetime and parental youth and adulthood
Table 2. Partial output for the association between the minimum and maximum temperature and reaction norms (change in calorie uptake for a two-degree increase in mean temperature) in multivariate models that also included information on altitude (proportion of the population residing below 500 m asl, height (m), economic conditions (four categories in 2017), and daily calorie intake (calories person−1 day−1) for 80 countries.
Table 2. Partial output for the association between the minimum and maximum temperature and reaction norms (change in calorie uptake for a two-degree increase in mean temperature) in multivariate models that also included information on altitude (proportion of the population residing below 500 m asl, height (m), economic conditions (four categories in 2017), and daily calorie intake (calories person−1 day−1) for 80 countries.
ModelParameter
(Temperatures)		EstimateSEt Valuep-ValueR2
1Minimum−1.062.04−0.520.610.17
Maximum7.704.111.870.07
5Minimum−2.244.88−0.460.650.20
Maximum22.359.822.280.03
10Minimum0.267.920.030.970.20
Maximum19.5315.951.220.23
20Minimum11.7814.320.820.410.11
Maximum−25.3228.82−0.880.38
30Minimum9.2918.100.510.610.04
Maximum−21.9236.44−0.600.55
1010Minimum3.429.720.350.730.06
Maximum−6.8319.58−0.350.73
2020Minimum−29.1815.51−1.880.060.18
Maximum47.4831.231.520.13
3030Minimum−50.2416.99−2.960.0040.37
Maximum127.9234.193.740.0004
D10Minimum5.2713.480.390.690.13
Maximum−9.7827.13−0.360.72
D20Minimum−16.3623.14−0.710.480.10
Maximum34.9146.580.750.46
D30Minimum−56.3930.38−1.860.070.17
Maximum164.2361.162.690.009
30
within D30		Minimum−17.1915.56−1.100.270.07
Maximum41.8331.331.340.19
3030
within D30		Minimum−39.2018.19−2.150.030.24
Maximum122.4036.623.340.001
Table 3. Statistical output for analysis of Model 3030 mean reaction norms and AIC values (n = 80) for an association with altitude temperatures, (proportion of the population residing below 500 m asl, height (m), economic conditions (four categories), and daily calorie intake (calories person−1 day−1).
Table 3. Statistical output for analysis of Model 3030 mean reaction norms and AIC values (n = 80) for an association with altitude temperatures, (proportion of the population residing below 500 m asl, height (m), economic conditions (four categories), and daily calorie intake (calories person−1 day−1).
VariableParameterEstimateSEt-ValuePr > |t|R2
Mean temperature Reaction normIntercept10,437.596703.921.560.120.37
Minimum temperature−50.2416.99−2.960.004
Maximum temperature127.9234.193.740.0004
Low-altitude populations−4.524.69−0.960.34
Height−84.9641.13−2.070.04
High income1422.31569.692.500.02
Upper middle income1403.03424.273.310.002
Lower middle income51.19384.510.130.89
Low income0.00
Daily calorie intake0.230.490.470.64
Relative AIC valueIntercept0.95880.21804.4<0.00010.11
Minimum temperature−0.00010.0006−0.10.92
Maximum temperature−0.00150.0011−1.370.18
Low-altitude populations0.00030.00022.160.03
Height0.00060.00130.430.66
High income−0.01770.0185−0.950.34
Upper middle income−0.00730.0138−0.530.60
Lower middle income−0.00120.0125−0.10.92
Low income0
Daily calorie intake−1.7 × 10−51.6 × 10−5−1.060.29
Table 4. Statistical output for analysis of model 3030, three reaction norms (n = 80) for an association within multivariate models without other state variables.
Table 4. Statistical output for analysis of model 3030, three reaction norms (n = 80) for an association within multivariate models without other state variables.
VariableParametersEstimateSEt-ValuePr > |t|R2
Mean temperature Reaction normIntercept−1766.52831.69−2.120.0370.13
Minimum temperature−54.4816.38−3.330.001
Maximum temperature101.4437.482.710.008
Maximum Temperature
Reaction norm		Intercept−1638.26836.67−1.960.0540.13
Minimum temperature−53.5416.48−3.250.002
Maximum temperature100.0037.702.650.009
Minimum Temperature
Reaction norm 		Intercept−128.26776.71−0.170.870.00
Minimum temperature−0.9415.30−0.060.95
Maximum temperature1.4435.000.040.97
Table 5. Output from a cross-sectional analysis of calorie day−1 person−1 intake in 162 populations by a multivariate model that included mean, maximum, and minimum temperatures, altitude (proportion of the population residing below 500 m asl, height (m), and economic conditions (four categories). Note that the 162 countries include a wider range in temperatures and temperature amplitudes than the eighty countries analysed in Table 2, Table 3 and Table 4. Additionally, the sum of the estimates for minimum and maximum temperature almost matches the estimate for mean temperature.
Table 5. Output from a cross-sectional analysis of calorie day−1 person−1 intake in 162 populations by a multivariate model that included mean, maximum, and minimum temperatures, altitude (proportion of the population residing below 500 m asl, height (m), and economic conditions (four categories). Note that the 162 countries include a wider range in temperatures and temperature amplitudes than the eighty countries analysed in Table 2, Table 3 and Table 4. Additionally, the sum of the estimates for minimum and maximum temperature almost matches the estimate for mean temperature.
ParameterEstimateSEt ValuePr > |t|R2
Intercept2049.7339.56.04<0.00010.75
Mean temperature−244.658.3−4.19<0.0001
Maximum temperature105.332.93.20.002
Mean × maximum temp.1.60.81.980.04
Minimum temperature121.726.94.53<0.0001
Mean × minimum temp.−1.60.4−4.27<0.0001
Median Age13.56.52.090.04
Low-altitude populations2.00.82.320.02
High income552.5102.45.39<0.0001
Upper middle income447.273.16.11<0.0001
Lower middle income211.664.33.290.001
Low income0.0
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Jensen, P.M.; Sørensen, M. Moderate Impact of Increasing Temperatures on Food Intake in Human Populations. Challenges 2025, 16, 34. https://doi.org/10.3390/challe16030034

AMA Style

Jensen PM, Sørensen M. Moderate Impact of Increasing Temperatures on Food Intake in Human Populations. Challenges. 2025; 16(3):34. https://doi.org/10.3390/challe16030034

Chicago/Turabian Style

Jensen, Per M., and Marten Sørensen. 2025. "Moderate Impact of Increasing Temperatures on Food Intake in Human Populations" Challenges 16, no. 3: 34. https://doi.org/10.3390/challe16030034

APA Style

Jensen, P. M., & Sørensen, M. (2025). Moderate Impact of Increasing Temperatures on Food Intake in Human Populations. Challenges, 16(3), 34. https://doi.org/10.3390/challe16030034

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop