Brain size evolution in hominins constitutes a crucial evolutionary trend, yet the underlying mechanisms behind those changes are not well understood. Here, climate change is considered as an environmental factor using multiple paleoclimate records testing temperature, humidity, and precipitation against changes to brain size in 298 Homo specimens over the past fifty thousand years. Across regional and global paleoclimate records, brain size in Homo averaged significantly lower during periods of climate warming as compared to cooler periods. Geological epochs displayed similar patterns, with Holocene warming periods comprising significantly smaller brained individuals as compared to those living during glacial periods at the end of the Late Pleistocene. Testing spatiotemporal patterns, the adaptive response appears to have started roughly fifteen thousand years ago and may persist into modern times. To a smaller degree, humidity and precipitation levels were also predictive of brain size, with arid periods associated with greater brain size in Homo. The findings suggest an adaptive response to climate change in human brain size that is driven by natural selection in response to environmental stress.
Understanding evolutionary brain size changes within a genus is critical to determining how different species adapt to broader environmental and cultural changes. This is particularly true for humans, given the reliance on increases in cognitive capacity over the past two million years. Yet we know surprisingly little about what causes adaptive changes to the human brain [Dunbar and Shultz, 2017].
Increasing brain size has been one of the more consistent by-products of hominin evolution [Pilbeam and Gould, 1974; Beals et al., 1984; Ruff et al., 1997; McHenry and Coffing, 2000; Rightmire, 2004; Shultz et al., 2012; Antón et al., 2014; Du et al., 2018; González-Forero and Gardner, 2018]. Early Homo benefited from large increases in relative and absolute brain size [Pilbeam and Gould, 1974; McHenry and Coffing, 2000; Antón et al., 2014; Püschel et al., 2021; Gingerich, 2022]. This was followed by new species within the genus, including H. erectus, H. heidelbergensis, and H. neanderthalensis, all of whose speciation events were marked in part by increased brain size and encephalization [Ruff et al., 1997; McHenry and Coffing, 2000; Antón et al., 2014]. H. sapiens evolved even greater brain size and encephalization around their speciation event roughly 300,000 years ago [Rightmire, 2004; Antón et al., 2014; Hublin et al., 2015; Du et al., 2018].
While the evolutionarily benefits of Homo brain growth are well known – increased cognitive capacity, tool use, language acquisition, sociality [Jerison, 1973; Dunbar, 2003; Shultz et al., 2012; Antón et al., 2014; González-Forero and Gardner, 2018] – the reasons why the human brain increased over time are not well understood [Dunbar and Shultz, 2017]. Numerous hypotheses have been proposed to explain Homo brain size increases, including environmental and ecological changes [Antón et al., 2014; González-Forero and Gardner, 2018], diet and food availability [Aiello and Wheeler, 1995; Antón et al., 2014], competitive [Bailey and Geary, 2009] and cooperative [Dunbar, 2003; Antón et al., 2014] dynamics within the genus as well as with other animals [González-Forero and Gardner, 2018], and as a general response to greater cognitive demands [Dunbar and Shultz, 2017]. It is likely that some combination of each of these factors, in addition to other variables, played a role in driving brain size increases in Homo.
With respect to environmental factors, variability in precipitation levels [Will et al., 2021], net primary production [Stewart, 2014; Will et al., 2021], climate pulses [Shultz and Maslin, 2013], and climate variability [Ash and Gallup Jr, 2007; Antón et al., 2014] have all been proposed to impact brain size to some extent. However, rigorous testing of these hypotheses has not been conducted, and as such, the mechanisms driving brain size changes are not fully understood. Prior studies have relied on data records that do not account for geochronological errors that are prevalent in older fossils and do not take into account the spatiotemporal lag associated with adaptive responses to environmental variables [Sabeti et al., 2006; Dehasque et al., 2020; Evershed et al., 2022]. More importantly, no study to date has controlled for body size and encephalization, factors known to contribute to brain size variability [Jerison, 1973]. As a result, studies testing evolutionary brain size changes [Ash and Gallup Jr, 2007; Bailey and Geary, 2009; Shultz and Maslin, 2013; Stewart, 2014; Will et al., 2021] have either been inconclusive [Bailey and Geary, 2009; Will et al., 2021] or contested [Shultz et al., 2012].
At greater issue is that almost all prior studies have generally assumed brain size has persistently increased throughout time. However, despite general trends of increasing brain size in Homo across time, Homo brain size has not been on a persistent upward march. Periods during the early Middle Pleistocene (roughly 875,000 years ago) [Ruff et al., 1997] and the Holocene (roughly 12,000 years ago to present) [Henneberg, 1988; Henneberg and Steyn, 1993; DeSilva et al., 2021; Stibel, 2021] both reflected deceleration in brain size, and modern evidence shows a decrease in both brain size and encephalization [Stibel, 2021]. For humans, there is a clear adaptive advantage to larger brains, and speciation events have repeatedly yielded lineages with greater relative brain size; increases in cognitive ability have consistently outweighed the costs of expensive brain tissue. As such, it is less clear why the brain would shrink during certain evolutionary periods, leading to hypotheses that the cognitive advantage that comes with larger brains has a breakpoint where it may no longer be evolutionarily beneficial [DeSilva et al., 2021]. There are exceptions, however, to the brain size advantage: brains are metabolically expensive and produce high levels of heat relative to their mass [Martin, 1981; Aiello and Wheeler, 1995; Hublin et al., 2015; Dunbar and Shultz, 2017]. Environmental changes, as a result, can be a key determinant of the net value of greater brain size. This is particularly true with regards to changing climates, as climate has been shown to directly impact metabolic and thermoregulatory systems in animals [Bergmann, 1847; Beals et al., 1984; Sheridan and Bickford, 2011; Kasabova and Holliday, 2015; Pomeroy et al., 2021; Will et al., 2021]. Given recent global warming trends, it is critical to understand the impact of climate change, if any, on human brain size and ultimately human behavior.
Materials and Methods
The challenge with testing Homo brain size changes is a lack of comparable sample data. Few intact skeletons exist, and brain size estimates are difficult to ascertain when derived from cranial capacity leveraging fragmentary skeletal remains. What measurements are available tend to be sparse and come from small samples without sufficient information about the specimen. In addition, precise carbon-14 dating is not available beyond roughly 50,000 years which makes it difficult to ascertain precise geochronology of older fossils. Prior studies [Ash and Gallup Jr, 2007; Bailey and Geary, 2009; Shultz and Maslin, 2013; Stewart, 2014; Will et al., 2021] suffered as a result of either insufficient sample breadth or an inability to control for dating errors, body size, taxon, and encephalization, all of which have been shown to impact brain size [Jerison, 1973]. The present study utilized a relatively large sample (n = 298) which allowed for controls of each of the above confounds. Given that body size is highly correlated with brain size in Homo [Pilbeam and Gould, 1974; Beals et al., 1984; Ruff et al., 1997; McHenry and Coffing, 2000; Grabowski, 2016; Stibel, 2021], latitude and sex – two strong correlates of body size – were used as proxies for body size and served to control for encephalization differences [Jerison, 1973; Beals et al., 1984; Ruff et al., 1997; Ruff et al., 2018]. Fossils were also limited to the past 50,000 years (50 kyr) to control for geochronology errors and because it is a period of relative stasis in brain/body isometry [Ruff et al., 1997; McHenry and Coffing, 2000; Stibel, 2021]. While limiting the study to the past 50 kyr presents some difficulties in analyzing longer term trends, there is a particular advantage in that the climatic period can be broken down into two dramatic temperature periods before and after the last interglacial, roughly 17 kyr BP.
Ten studies that documented cranial capacity (cm3) in Homo over the past 50 kyr BP were used to compile cranial data (see Dataset 1 for sources of measurements and dating). Cranial measurements were obtained from meta-analyses, published sources, and relevant updated data [Holloway, 1980; Holloway, 1981; Beals et al., 1984; Brown, 1992; Henneberg and Steyn, 1993; Ruff et al., 1997; De Miguel and Henneberg, 2001; Hawks and Wolfpoff, 2001; Manjunath, 2002; Lordkipanidze et al., 2013]. Incomplete measures, unusable skull fragment data, and juvenile specimens were discarded. Where multiple measurements were available for a particular skull, they were aggregated to provide an average cranial capacity estimate. In total, 373 independent cranial capacity measurements were utilized across 298 skulls. Brain size estimates were derived from cranial capacity measurements, as cranial volume has been shown to highly correlate with brain size in both modern and prehistoric humans [Pilbeam and Gould, 1974; Beals et al., 1984; McHenry and Coffing, 2000]. Cranial capacity was converted to brain size using a formula derived from a least-squares regression (r2 = 0.995, LSR) of 27 primate species (brain mass = 1.147 x cranial capacity0.976) [Ruff et al., 1997]. Where precise dates or carbon-14 dating were available (n = 266), that source was used; otherwise, the average was taken across all date estimates (n = 32). Specifics on the type of measure, chronometric date, sex, taxon, geography, formula derivations, and all sources are available in Dataset S1 and Stibel, 2022.
Brain size data were compared to four climate records. The primary dataset utilized paleotemperatures derived from the European Project for Ice Coring in Antarctica (EPICA) at Dome C [Jouzel et al., 2007] (online suppl. Dataset S1; for all online suppl. material, see www.karger.com/doi/10.1159/000528710). Records from the ice core at EPICA Dome C provide precise 100-year surface temperatures that have been found to correlate closely with stacked climate data across 58 different regional sites, including regions near the fossils used in this study [Lisiecki and Raymo, 2005; Jouzel et al., 2007]. Brain size estimates were compared to coeval temperatures as well as to past temperatures at intervals dating back 5 kyr, 10 kyr, and 15 kyr from the fossil date to control for the uncertainty surrounding the delayed adaptive response of the spatiotemporal interaction between the two variables. To test for evolutionary effects and control for the accuracy of carbon dating which is not reliable within 100-year periods [Wright, 2017], the temperature data were also cataloged for the present study into increments of 5,000, 10,000, and 15,000 years (online suppl. Dataset S1).
The ice core at EPICA Dome C provides precise surface temperature measurements dating back 809,950 years [Jouzel et al., 2007]. Temperature records at EPICA Dome C were measured as a function of differences in past temperatures (∆T) relative to the most recent 100-year temperature average. Temperature change was calculated against a modern mean temperature as follows: “The surface temperature change, ∆Ts (estimated with respect to the mean value calculated over the last millennium), (was) calculated using the present-day spatial slope of 6.04‰/°C after correction for the change in the isotopic composition of the ocean (and) a slight correction linked with changes in the altitude of the ice sheet as calculated from the glaciological model used to derive the timescale” [Jouzel et al., 2007]. Average temperature change (mean ∆T) across the EPICA Dome C records as compared to the last millennium was calculated at −5.30°C.
The global climate records also offer insight into regional differences. The climate records produced from Dome C in Antarctica have been shown to correlate closely with stacked benthic δ18O climate data derived from oceanic sediment cores across 58 different regional sites, including sites located near the fossil remains used in this study [Lisiecki and Raymo, 2005; Jouzel et al., 2007]. The stacked regional data displayed strong correlations (average r2 of 0.88 after alignment) with the Antarctic average [Jouzel et al., 2007], suggesting relative consistency of the ice sheet record and stacked regional averages. An additional measure using regional surface temperatures derived from a sediment core from Lake Malawi in East Africa [Johnson et al., 2016] (see below for more details) was also compared to the EPICA Dome C record and was found to be highly correlated during the periods used herein over the past 50 kyr (LSR, r2 = 0.78, p < 0.0001, ANOVA).
Surface temperatures (°C) derived from sediment samples recovered from Lake Malawi (10°–14° S in eastern Africa) over the past 1.3 million years [Johnson et al., 2016] were used as proxies for regional temperatures in Africa (online suppl. Dataset S1). The Lake Malawi record was shown to correlate with past changes in atmospheric carbon dioxide and, when compared to average rainfall estimates, demonstrated a relationship between warmer, wetter periods and cooler, drier cycles [Johnson et al., 2016].
Grain-size analyses of siliciclastic marine sediments from the coast of Mauritania (core GeoB7920; 2,278-meter water depth) were utilized to derive a humidity index (mm) for Northwest Africa starting at 500 years ago and going through the entire 50 kyr BP period (online suppl. Dataset S1) [Tjallingii et al., 2008]. To derive relative changes in humidity, log ratios of three core endmembers from the marine sediment were utilized (log (hemi-pelagic mud)/(coarse aeolian dust + fine aeolian dust)), whereas hemi-pelagic mud provided a measure of river runoff and aeolian dust records, which served as a proxy for subaerial erosion and continental vegetation cover [Tjallingii et al., 2008].
Pollen sequences (mm) were used to derive a time series of mean annual precipitation level estimates by calculating the difference between the reconstructed absolute value of the pollen counts and modern precipitation values (online suppl. Dataset S1) [Bonnefille and Chalie, 2000]. Four African pollen sequences were derived from two sources dating back to 40 kyr BP: three samples from the equatorial mountains in the Burundi highlands at 1,850–2,240 m of elevation and a fourth sample from the South Tanganyika basin at 770 m of elevation (the fourth sample was derived from a prior record provided by the same researcher and reconstructed with a new synthesis in conformance with updated methodology) [Bonnefille and Chalie, 2000]. Six additional sequences were available but not utilized as there were no records with matching time periods to available fossil remains.
Paleoclimate records were provided over 100-year averages or greater, whereas fossil records were dated to the year, as best estimated by geochronologists without regard for variance. The temperature record was rounded down to match the estimated age of each skeletal remain. Given the considerable variability in fossil dating, efforts were made to ensure that the results accounted for potential geochronological errors. Homo records in the sample were limited to the past 50,000 years, where precise records or radiometric carbon dating accounted for the majority of date estimates (266 of 298 specimens). Where exact dates or radiometric dating were not available, multiple estimates were used and averaged. Additionally, the fossils were clustered into 100-year, 5,000-year, 10,000-year, and 15,000-year groupings, the latter of which can account for dating errors up to 15,000 years. The full dataset is available in the online supplementary Data and Stibel, 2022.
A significant relationship was found between temperatures at EPICA Dome C and brain size estimates over the past 50 kyr (Fig. 1a) (p < 0.0001, ANOVA), which held constant after controlling for geography, sex, and taxon (p < 0.0001, ANCOVA). Despite the correlation, it is important to note that the climate and brain size records do not appear to correspond temporally (Fig. 1b–c). Brain size changes appear to take place thousands of years after changes to climate, and this is particularly pronounced after the last glacial maximum, approximately 17 kyr. While acclamation unfolds within a single generation and natural selection can happen in as short as a few successive generations, species level adaptation often takes many successive generations [Sabeti et al., 2006; Evershed et al., 2022]. The time series trends demonstrate what appears to be a pronounced delay that may have been driven by evolutionary adaption (Fig. 1d–f, online suppl. S1).
To test for evolutionary effects, brain size changes were compared to prior generational climate periods as well as across temperatures averaged over extended periods of time. Brain size estimates were first modeled against 100-year average temperatures dating back 5 kyr, 10 kyr, and 15 kyr (online suppl. Fig. S2). A second set of models was derived for brain size estimates as compared to temperatures averaged across 5 kyr, 10 kyr, and 15 kyr (online suppl. Fig. S2). Across each model, brain size remained significantly correlated with global average temperatures (all tests, p < 0.0001, ANOVA, online suppl. Table S1). Inclusion of sex and latitude as covariates increased the explanatory power of the models to approximately 40% (all tests, p < 0.0001, ANCOVA, online suppl. Table S2), with the 10 kyr spatiotemporal lag showing the highest predictive power at roughly 42% (online suppl. Fig. S2).
To control for geochronological errors and to increase statistical power, the Homo sample was next divided into groups of individuals living during periods that were above and below average temperatures. To account for a bias toward warmer climatic periods in the skeletal sample, the data were compared using the average temperatures derived solely from the 298 periods where skeletal remains were available (mean ∆T of −1.07°C). Across 100-year mean average global temperatures over the past 50,000 years [Jouzel et al., 2007], brain size in Homo specimens living during cooler than average climatic periods was found to be significantly larger than that of those living during warmer periods (p < 0.0001, t test). Homo brain size during cooler periods averaged 1,426.31 g ± 137.30 (mean ± standard deviation) (n = 65) as compared to 1,280.89 g ± 141.67 (n = 233) for warmer periods, or a roughly 10.74% difference (Fig. 2a; Table 1).
To control for evolutionary effects, the brain size data were also grouped using data aggregated into 10,000-year increments (while 10 kyr increments were chosen for representation in the study, tests using temperatures averaged over 5-kyr and 15-kyr periods yielded directionally similar results). Similar to the 100-year data, Homo brain size was found to be significantly larger during periods of global cooling (1,421.69 g ± 136.56 [n = 70]) as compared to warmer cycles (1,279.12 g ± 141.77 [n = 228]), or a roughly 10.57% difference (p < 0.0001, t test) (Fig. 2b; Table 1).
Body size in both modern and prehistoric Homo tends to be larger in high latitude (above 30° North) geographies [Beals et al., 1984; Ruff et al., 1997]. Given that body size tends to correlate highly with brain size [Pilbeam and Gould, 1974; Beals et al., 1984; Ruff et al., 1997; McHenry and Coffing, 2000; Grabowski, 2016; Stibel, 2021], the significant changes found in brain size over time could have been a result of Homo dispersal across latitudinal clines. Despite a bias toward high-latitude Homo in the sample (220 of 298), controlling for body size across clines above and below 30ºN latitude did not materially change the results, and significant differences persisted in brain size across climatic periods (p < 0.0001, ANCOVA, across 100- and 10,000-year periods). During cooler average temperatures, larger bodied Homo brain size was 8.14% larger (p < 0.0001, t test) and smaller bodied Homo brain size was 19.87% larger (p = 0.0001, t test) than during warmer than average climatic cycles (Fig. 3a, b; Table 1).
Similar to the ecogeographic patterns, sexual dimorphism is present in Homo, with males on average having larger body sizes than females [Ruff et al., 1997; Ruff et al., 2018]. Despite a bias toward males in the sample (167 of 257 sexed specimens), controlling for body size through sex did not change the relationship between climate change and brain size (p < 0.0001, ANCOVA, across 100- and 10,000-year periods). Using sex as an intermediate control, larger bodied Homo averaged 8.40% larger brain size (p < 0.0001, t test) during cooler average temperatures as compared to warmer periods (smaller bodied Homo was directionally consistent – 4.89% greater brain size during cooler than average periods – but not significant, possibly due to limited sample size, with only 11 cooler than average specimens) (Fig. 3a, b; Table 1).
Because the sample consisted of modern and archaic Homo, robusticity differences across species could also have influenced the results. Anatomically modern H. sapiens (AM Homo) in particular tend to be significantly more gracile with less skeletal and body mass than both earlier hominins (Middle Pleistocene and archaic Homo) and more recent archaic H. sapiens (i.e., Neanderthals) [Hawks and Wolfpoff, 2001; Harvati et al., 2006; Hublin, 2009], which could account for some of the effect of climate on brain size given the estimates rely on skeletal measurements. As compared to the archaic fossils in the sample, brain size was 9.96% smaller for AM Homo on average (p < 0.005, t test). To control for robusticity differences, the impact of climate change was tested excluding the archaic Homo fossils, which eliminated 9 of the 298 specimens. Across AM Homo, the differences in brain size during colder and warmer periods persisted (p < 0.0001, t test). During colder average periods (∆Ts below −0.86°C), brain size across AM Homo averaged roughly 11.02% larger than during warmer than average periods, or roughly 1,422.03 g ± 142.52 (n = 56) as compared to 1,280.89 g ± 141.67 (n = 233). The results held directionally within 10,000-year periods and after controlling for body size (all results significant [p < 0.01, t test], except for small bodied sexed specimens, which were directionally similar but not significant) (Fig. 3a, b; Table 1).
The results present a general pattern of changing brain size in Homo that is correlated with climate change as temperatures increase and decrease. On closer inspection, the effects of climate change on brain size appear more pronounced nearest the mean and do not accelerate as temperatures grow more extreme. While significant differences were found for moderate temperature ranges (interquartile range, p < 0.0001, t test), there were no significant differences found in brain size as temperatures became more extreme (all tests, p > 0.50, t tests). As with the broader Homo sample, AM Homo brain size also appeared more sensitive to moderate temperature changes with an adaptive response that occurred in advance of extreme temperatures (interquartile range, p < 0.001, t test; extreme temperatures, all tests p > 0.50, t test) (Fig. 4).
Temporal periods produced similar patterns of changes to brain size. Testing temporal periods within the past 50,000 years presents a unique opportunity, as the time period includes the last interglacial that produced consistently colder average temperatures up until the end of the Late Pleistocene (mean ∆T of −6.96°C) and the Holocene (mean ∆T of 0.51°C), which has produced significantly warmer temperatures on average throughout present day (p < 0.0001, t test). The Holocene is a period noted for declining brain size generally, and numerous hypotheses have been proposed to explain the temporal decline [Henneberg, 1988; Henneberg and Steyn, 1993; DeSilva et al., 2021; Stibel, 2021], but global warming has not been directly tested against Homo brain size declines. Testing the two epochs independently produced effects consistent with the interquartile results. Whereas brain size in Homo during the late Late Pleistocene (50–12 kyr BP) was significantly larger than during the Holocene (12 kyr BP-present) (p < 0.0001, t test), there were no significant differences between colder and warmer periods within each epoch (all results p > 0.50, t test), suggesting that most of the changes in brain size in Homo happened as a result of the deglaciation period. This was further represented by the temporal trends in brain size and climate change (Fig. 1, online suppl. S1). Pleistocene brain size averaged 1,426.96 g ± 139.28 (n = 63) as compared to 1,281.95 g ± 141.58 (n = 235) during the Holocene, a roughly 10.71% difference (Fig. 5).
Because Homo dispersal originated out of Africa, three regional African climate records were also tested against the brain size sample: sedimentary records, marine sediment layers, and pollen sequences (online suppl. Dataset S1). Sedimentary records have been shown to be predictive of regional temperatures [Johnson et al., 2016], and the sample utilized was highly correlated with the EPICA Dome C climate record (LSR, r2 = 0.78, p < 0.0001, ANOVA); marine sediment layers have been shown to be predictive of humidity levels and vegetation cover [Tjallingii et al., 2008]; and pollen sequences have been shown to be predictive of precipitation levels as well as estimated rainfall and water availability [Bonnefille and Chalie, 2000] (See Methods for details). Similar to the global temperature results, colder regional temperatures were predictive of larger brain size in humans; in addition, more arid conditions (lower average rainfall and humidity) were found to be predictive of larger brain size in humans (all tests, p < 0.05, ANCOVA controlling for sex and latitude across Homo and AM Homo). Differences in Homo brain size were roughly 6.15% for regional temperature (p < 0.0001, t test), 5.28% for humidity (p < 0.002, t test), and 2.74% for precipitation levels (nonsignificant, p = 0.061, t test) (Table 1).
For each of the above regional climate records, the tests were also performed limiting the Homo sample to the continent of Africa (n = 26) and sub-Saharan Africa (n = 23) (there were no archaic specimens, so the entire sample consisted of AM Homo). In all cases, the patterns were directionally consistent with the overall findings, but no significant results were produced through group testing (all results, ≥0.06, t test), possibly as a result of the low sample size after limiting to a single geography. Modeling the African temperature record, humidity estimates, and precipitation levels against African Homo and sub-Saharan African Homo accounted for roughly 42% (LSR, r2 = 0.421, p < 0.01, ANOVA) and 44% (LSR, r2 = 0.439, p < 0.01, ANOVA) of the variability in brain size, respectively.
The reasons for hominin brain size adaptions are complex and not well understood. Climate has frequently been proposed as a driver of brain size evolution [Potts, 1996; Ash and Gallup Jr, 2007; Shultz and Maslin, 2013; Antón et al., 2014; Stewart, 2014; Will et al., 2021], but testing has not produced conclusive results [Ash and Gallup Jr, 2007; Bailey and Geary, 2009; Shultz et al., 2012; Shultz and Maslin, 2013; Stewart, 2014; Will et al., 2021]. By systematically controlling for encephalization, body size, and taxon, the present study was able to isolate evolutionary changes to the brain and systematically test the impact of climate change. The results suggest that climate change is predictive of Homo brain size, and certain evolutionary changes to the brain may be a response to environmental stress.
Adaptive responses to natural selection are difficult to unpack given the spatiotemporal delay between the underlying cause and effect [Sabeti et al., 2006; Burger et al., 2020; Dehasque et al., 2020; Evershed et al., 2022]. While macroevolutionary adaptations can take millions of years, some modern adaptive responses in Homo, such as lactose tolerance, have evolved more quickly [Gerbault et al., 2011; Burger et al., 2020; Evershed et al., 2022]. Given the data within this study can only provide correlational support for spatiotemporal relationships, future studies will be needed to confirm any assumptions regarding the responsiveness of brain size to climate change. However, the results of the current study demonstrate what may be an evolutionary adaptation to environmental stress in human brain size beginning roughly 15,000 years ago and persisting through present day.
The primary variable tested herein was global temperature change. There is a strong basis for temperature differences impacting brain size but that has not previously been demonstrated evolutionarily. To maintain thermoregulatory balance in response to different temperatures, physical size is believed to be adapted to regional geographies given that smaller sizes dissipate heat better than larger volumes with less relative surface area [Bergmann, 1847; Beals et al., 1984; Sheridan and Bickford, 2011; Kasabova and Holliday, 2015]. This is true for human body size overall and brain size [Beals et al., 1984]. Temperature-driven climate change has also been shown to impact body size evolutionarily [Will et al., 2021; Pomeroy et al., 2021; Stibel, 2021]; however, the effect has not previously been demonstrated for brain size. The present study demonstrates broad brain size changes in response to changing temperatures using both a global paleoclimate record, as well as a regional climate record from Africa. The impact of climate change on human brain size appears to be similar to how geographic temperature differences affect body size in response to thermal stress: global warming tends to favor smaller brain size, whereas global cooling favors larger size.
In addition to temperature changes, humidity and precipitation were also associated with brain size adaptations, although this effect was less pronounced. In both cases, aridity was predictive of greater brain size and wetter periods were predictive of smaller size. As with thermal stress producing larger brains in cooler climates, it appears as if environmental stress related to decreased water availability may produce larger brain size in Homo. Some caution should be considered in interpreting these results in at least two respects: first, climate records that measure water production tend to be highly localized, and the data used herein were limited to regions in Africa; and second, at least with respect to certain African climates, arid conditions tend to correlate with cooler climates [Johnson et al., 2016], such that the impact of aridity may be mediated by the larger effect of temperature on brain size.
Environmental hypotheses that have been proposed for brain size increases can broadly be summarized into two groups: those that assume hospitable environmental conditions (wetter, fertile, and more consistent climates) increase food availability which allows for larger brain size [Stewart, 2014; Will et al., 2021] and those that assume climate stressors (cold or variable temperatures and arid environments) produce brain size increases as a coping response to more extreme conditions [Potts, 1996; Ash and Gallup Jr, 2007; Antón et al., 2014; Will et al., 2021]. With respect to increasing brain size, the results herein generally support the latter hypothesis and are incongruent with the first. There was no relationship found between hospitable environments and greater brain size. Rather, there was evidence that colder, more arid periods correspond to larger brain size.
In its response to temperature changes and water availability, the human brain appears to be reacting to environmental stress as opposed to another adaptive mechanism. However, when taking into account brain size declines, climate stability appears to be the more critical environmental variable underlying brain size changes. While past hypotheses have focused on brain size increases, the present data demonstrate a more general response to climatic factors wherein brain size modulates depending on the environmental conditions. Brain size was larger during periods of colder and more arid environments and was smaller when temperatures were warmer and wet. This was found irrespective of time period, and the latest Holocene warming cycle was predictive of brain sizes in Homo that were smaller than prior periods. The changes in brain size were proximate to the mean average temperatures, and the general trend toward larger or smaller brain size did not accelerate as climates grew more extreme, suggesting a pattern of changing brain size limited to transitionary periods as climates become more extreme. However, some caution should be taken when considering more granular trends given fossil dating constraints and spatiotemporal delays, so it is best to interpret the data as demonstrative of a broader effect across warming and cooling cycles until more precise methods are available.
These findings have other taxonomic and phylogenic implications. Specifically, some variation in brain size within the genus Homo may be attributable to climate in addition to broader evolutionary causes and should be considered when interpreting observed morphological variation. Speciation events have often been marked in part by changes to brain size in hominins, and those decisions may be worth reconsidering in the context of the present results. Cranial morphology, brain mass, and encephalization in particular have been considered as factors for classification within the genus Homo or as criteria for demarcating a new species [Wood and Collard, 1999; Harvati et al., 2006; Hublin, 2009; Hublin et al., 2015], but these variables may be affected by climate change. By way of example, there has been considerable debate as to whether H. neanderthalensis should be considered a sub-species of H. sapiens, in part because of their large brain and body mass [Hawks and Wolfpoff, 2001; Harvati et al., 2006; Hublin, 2009; Hublin et al., 2015; VanSickle et al., 2020]. Neanderthal remains have almost exclusively been found in high-latitude regions and from cooler climatic periods, both of which are consistent with bigger body and brain sizes [Hawks and Wolfpoff, 2001; Harvati et al., 2006; Hublin, 2009]. The differences in brain and body size may be more a function of climate than taxon, and the present results may offer some additional guidance on whether H. neanderthalensis should remain independent of the H. sapiens clade. At a minimum, it may be worth considering whether climatic periods should act as a control when considering morphological differences within the genus.
The effects of climate change on brain size may have been intermediated by body size adaptations, as similar trends have been found for Homo body size changes in response to climate change [Will et al., 2021; Pomeroy et al., 2021; Stibel, 2021]. Fossil records have shown that invertebrates, burrowing insects, and mammals all shrunk during past periods of global warming [Sheridan and Bickford, 2011; Peralta-Maraver and Rezende, 2021] and periods of global cooling have been shown to drive body size increases in many organisms [Millien et al., 2006]. These results are consistent with studies that have shown body size changes in response to climate change in Homo [Will et al., 2021; Pomeroy et al., 2021; Stibel, 2021]. It is not clear from the present study whether brain size was specifically selected for or against during periods of changing climate or whether phenotypic changes to the brain were a result of selective pressures acting on body size that caused brain size to drift alongside the body. Regression models have shown that selection on the brain or the body independently in organisms can impact both variables and produce a correlated response [e.g., Lande, 1979; Atchley, 1984; Riska and Atchley, 1985]. For hominins, at least one study has demonstrated a phenotypic link to brain size acting as the underlying variable under selection in both body and brain size changes during prehistoric periods [Grabowski, 2016]. In that case, brain size increases were determined to be the primary phenotype under selection, driving both brain and body size changes. However, the study did not contemplate decreases in size or climate as an intermediating variable. It is not clear as a result, whether the body, the brain, or a combination of the two are directly under selection in response to climate change as opposed to one trait drifting through pleiotropy as a result of the other. Given the human brain’s high thermal output and sensitivity to extreme heat [Falk, 1990], it may be that climate disproportionately effects brain size, but more data and analyses will be required to determine the interrelationship between brain and body size.
The underlying drivers of brain size changes are not fully clear. While the results demonstrate a relationship between climate change and Homo brain size, climate appears to account for only a small amount of the variation in brain size evolutionarily. Brain size adaptations are complex and are likely driven by other factors that affect the ecosystem (such as predation or other competitive dynamics), by an indirect consequence of a changing climate (such as vegetation levels and net primary production), or by means unrelated to climate (such as culture and technology). Environmental hospitability is a complex formula that includes many variables influencing physiological adaptation. While there is some evidence to support temperature as the primary climatic influence on Homo brain size, more work will be needed to determine whether the impact of climate change on Homo physiology is a result specifically of temperature changes or an indirect effect from other elements of a changing environment.
To the extent that temperature changes directly influence brain size, the present period of accelerated warming could lead to increased evolutionary pressure on the human brain. The earth is currently warmer than it has been at any time in the past 125,000 years, with temperature increases of more than 1.1°C during the past century and projections of an increase of at least 1.5°C by the end of the current century [Tollefson, 2021]. While this may be a sampling or dating error, there is some genetic evidence to suggest that brain size, general cognitive ability, and educational attainment are all being selected against in modern human populations [Reynolds et al., 1984; Bouchard Jr et al., 1990; Deary et al., 2007; Rietveld et al., 2013; Rietveld et al., 2014; Plomin and Deary, 2015; Sniekers et al., 2017; Okbay et al., 2016; Stibel, 2021]. Even a slight reduction in brain size across extant humans could materially impact our physiology in a manner that is not fully understood.
The author would like to thank Chris Ruff, Tim White, Robin Dunbar, and Jeremy DeSilva for a critical review of the manuscript; Dan Dennett and Phil Lieberman for helpful commentary on specific areas; Chris Ruff, Jean Jouzel, and Peter Brown for providing expanded datasets; Lindsey Long and Caitlin Mason for assistance with data collection, tabulation, and standardization; and Zack Stokes for assistance with mathematical equations and statistical modeling.
Statement of Ethics
An ethics statement was not required for this study type; no human or animal subjects or materials were used.
Conflict of Interest Statement
The author was employed by Bryant Stibel at the time of the publication and Bryant Stibel had no role in the study design; collection, analysis, and interpretation of data; writing of the report; or decision to publish.
No funding was acquired for this work.
Jeff Stibel was responsible for the results and write-up of the manuscript.
Data Availability Statement
The data supporting the findings of this study are available in the study by Stibel  and within the manuscript and its supplementary information files (online suppl. Data 1). Further inquiries can be directed to the corresponding author.