4  On the Latitude Hypothesis

The first mention of this hypothesis in English scientific literature dates back to at least 1973, as noted by Bohlen & Simpson (1973), with earlier hints of the idea from Erhard Haus and Franz Halberg in 1970 (Haus & Halberg, 1970, p. 101), building on discussions initiated by Jürgen Aschoff (Aschoff, 1969). Since then, numerous studies have explored this topic, yielding somewhat conflicting results (a systematic review is provided by Randler & Rahafar (2017)).

The hypothesis, also called environment hypothesis, posits that regions closer to the poles receive, on average, less annual sunlight compared to regions near the equator Figure 4.1. Consequently, regions around latitude 0° are thought to have a stronger solar zeitgeber. According to chronobiological theories, this stronger zeitgeber would enhance the synchronization of circadian rhythms with the light-dark cycle, resulting in lower variability and amplitude of circadian phenotypes. This reduced influence of individual endogenous periods is illustrated in Figure 4.2.

In contrast, populations near the poles experience a weaker solar zeitgeber, leading to greater variability and amplitude of circadian phenotypes. This disparity translates into differences in chronotype: equatorial populations tend to exhibit a morningness orientation, while populations at higher latitudes tend toward eveningness (Bohlen & Simpson, 1973; Roenneberg et al., 2003).

Figure 4.1: Annual changes in (a) twilight duration, (b) daylight hours, and (c) temperature across different latitudes. Each color shows a specific latitude, illustrating how these factors vary throughout the year.

Source: Reproduction from Hut et al. (2013).

Figure 4.2: Different chronotype distributions, influenced by strong and weak zeitgebers – black for strong and hatched for weak. An illustration of the effect hypothesized by the latitude hypothesis.

Source: Adapted from Roenneberg et al. (2003).

Some authors claim to found this association, but a closer look at the data reveals that it is not as clear as it seems. For example, Leocadio-Miguel et al. (2017) found a significant association between latitude and chronotype in a sample of \(12,884\) Brazilian participants. However, the effect size was negligible, with latitude explaining only about \(0.388\%\) of the variance in chronotype Figure 4.3. Considering the particular emphasis that the solar zeitgeber has on the entrainment of biological rhythms (as demonstrated in many experiments), it would not be reasonable to assume that the latitude hypothesis could be supported without at least a non-negligible effect size.

The findings of Leocadio-Miguel et al. (2017) are not consistent with the hypothesis that latitude is a strong predictor of chronotype, as the reported effect size is too small to be considered practically significant (Cohen, 1988). This highlights a common limitation of studies relying on Null Hypothesis Significance Testing (NHST) (Perezgonzalez, 2015). A p-value does not measure the effect size; instead, it represents the probability of observing the data (or something more extreme) assuming the null hypothesis is true, thus quantifying the likelihood of a type I error.

Figure 4.3: The mean scores (±SE) on the Horne & Östberg (HO) chronotype scale (Horne & Östberg, 1976) are presented across a latitudinal gradient, along with the corresponding annual average solar irradiation levels (W/m²). The HO scale comprises 19 items, with total scores ranging from 16 to 86; lower scores indicate a stronger evening orientation, while higher scores reflect a greater morning orientation. Notably, the y-axis exaggerates the visual impact of the differences, as it represents a range of only about 4.5 points, which may overstate the perceived significance of the effect.

Source: Reproduction from Leocadio-Miguel et al. (2017).

Several factors may invalidate this hypothesis, such as local clock time and social constraints (Skeldon & Dijk, 2021). To gain a more accurate understanding of the mechanisms underlying chronotype expression, it remains crucial to test this hypothesis in larger samples. This study aims to address that gap. In the following sections, the hypothesis will be tested using one of the largest chronotype datasets, to the author’s knowledge, with geocoding information integrated for a comprehensive analysis. The approach will adhere to sound statistical principles, incorporating a minimum effect size in the alternative hypothesis, as originally proposed by Neyman and Pearson (Neyman & Pearson, 1928a, 1928b).