Part 1: How to think about ketogenic diets within human evolutionary history
In the past decade ketogenic diets in humans have started to attract the attention of a few forward thinking researchers as well as a small number of online health enthusiasts. In any diet there are three main elements called macronutrients – fat, protein and carbohydrate. On a ketogenic diet most calories come from fat (65-90%), a moderate amount from protein (<10-25%) and a small amount from carbohydrate (0-15%).
A ketogenic diet is often mistaken for a high-protein diet. This is not accurate. A ketogenic diet means eating food that produces ketones, a kind of molecule in the blood that provides energy, like glucose does. Producing a high enough level of ketones is called being in ketosis and it is a metabolic state in which the body relies much less on glucose.
The who’s who of low-carbohydrate ketogenic research, headed by Accuros et al. in 2008 (1), defined ketogenic diets as containing <10% of calories from carbohydrates. There are two reasons that I prefer to give a range of 0-15%. First, scientists have not fed large populations in a controlled manner to see how much of each macronutrient is needed to shift more than half of them into nutritional ketosis (we lack empirical data on this). This is complicated by that fact that different people get into nutritional ketosis more or less easily because of various factors, like their level of insulin resistance for example. Second, scientists have not yet defined what the nutritional ketosis threshold is exactly, despite their being good approximations.
Before exploring the appropriateness of ketogenic diets for humans, I’d like to justify why I approach questions of human health and nutrition the way I do by introducing 2 concepts; evolution by natural selection and the consilience of evidence. It is with these in mind that I answer the question any self-respecting skeptic asks about the ketogenic diet or any other diet for that matter
Are ketogenic diets a fad or do they underpin something more substantive in human health?
Lets start with evolution by natural selection. As many of you might, I initially thought your average doctor, dietitian or nutritionist would know about the answer to the above question. Unfortunately, I realized the vast majority did not. Although the reasons for this deserve a longer discussion, the gist in my opinion is that unbeknownst to most of them, their training brushes over the most powerful idea in biology (maybe even in science!), namely natural selection.
This gets us to the heart of the matter. Proper scientific scrutiny of dietary or health practices cannot do without the lens of evolutionary biology. Richard Dawkins makes the general case for natural selection using what he calls the Explanation Ratio (2):
Assuming you are on board with the explanatory power of natural selection, we can now include this powerful tool in our scientific toolbox. It helps us do 2 important things. First, it helps us form hypotheses from simple principles, which is good scientific practice. Second, it allows us to filter out bad ideas, essentially throwing away those that do not fit with the theory of evolution (as best we understand it). It can ensure we fulfil the ‘first principle’ of science which famous physicist Richard Feynman summarized as
The first principle is that you must not fool yourself – and you are the easiest person to fool
Like any tool, evolution by natural selection is limited and is prone to misuse. We would do well to set ‘explanatory boundaries’ for it. Explanatory boundaries serve to outline the limits of how much an a framework or idea can tell us about something. For math heads, think of their purpose as analogous to a boundary value with additional constraints used in differential equations. In our instance of understanding diets from the evolutionary perspective, the question we can ask to help set our explanatory boundaries is:
What can natural selection tell us about ketogenic diets in humans?
The answer is that it can provide ‘ultimate’ explanations. Ultimate explanations are different from ‘proximate’ explanations. Proximate ones are given to us by clinical medical trials or biochemical experiments. Scott-Phillips et al. (3) define both terms as follows
Ultimate explanations are concerned with why a behavior exists, and proximate explanations are concerned with how it works. These two types of explanation are complementary and the distinction is critical to evolutionary explanation
Practically speaking, biochemistry and physiology tells us how ketosis (or ketogenesis) happens in humans (a proximate explanation) and evolutionary biology tells us why the state of ketosis was selected as a prominent metabolic feature of our species. Isn’t it cool, and useful, to know both the how and the why? Take the well-known problem of antibiotic resistance as an example. Molecular biologists can explain how antibiotic resistance happens by describing a chain of molecular events (the proximate cause of antibiotic resistance) but it takes a person steeped in the mechanics of natural selection to explain why bacteria developed this ability in the first place (the ultimate cause of antibiotic resistance).
Both the molecular biologist and the person steeped in evolutionary biology must derive their explanations from falsifiable and replicable experiments that test their claims. It’s a two-way street, where insights from one approach can serve to inspire or refine experiments in the other. Science is notoriously messy and a lot happens by serendipity. But we can get better at it the more we learn to ask both how and why. A note of caution is warranted at this point. Although the principles of natural selection are simple, there is a lot more to it than advantageous traits winning over disadvantageous ones. Consequently, a common fallacy arises, especially by those applying the evolutionary lens to nutrition, in the form of 2 claims:
We ate this in the past and so we should eat it nowadays
We did not eat this in the past and so we should not eat this nowadays
Both claims are incorrect. They are failed attempts at the reasonable question we should ask when thinking about nutrition:
How well or poorly adapted are we to the foods we eat nowadays?
When thinking about such a question I pull on threads from anthropology, clinical medicine, biochemistry, epidemiology and other fields. Similarly, evolutionary hypotheses often use evidence from independent and unrelated fields of study. When multiple lines of evidence converge towards a conclusion – even if individually weak – we call this consilience. Human health and nutrition are staggeringly complex, inexact sciences, lending themselves to an evolutionary framework bolstered by the consilience of evidence.
Moving on to the question of interest, let’s replace our original question of whether or not ketogenic diets are merely a fad with the more pointed question we just posed:
How well or poorly adapted are we to the foods we eat nowadays?
Keep the following thought in mind while we explore the answer. Any half-decent zookeeper asks themselves if their animals are well-adapted to the food and environment artificially provided to them. We humans are animals, and also deserve to have that same question asked of us – right?
Part 2: Are ketogenic diets evolutionarily appropriate for humans?
Lets start answering this question by first considering paleoanthropological data. There are observations made about what was eaten by hunter-gatherer groups still in existence in the last few hundred years. We can also look at clues left behind by the bones of Neanderthals and early modern Europeans. Additional clues come from
- what tools humans had and how we used them
- the fossils we left behind (including coprolites, literally fossilized poop!)
- the geography and climate of where human remains are found
- where humans fit in the tree of life relative to our hominidae kin
These are a few important examples of lines of evidence used by paleoanthropologists studying how humans became human (hominization). Many theories of hominization revolve around the influence of diet, the salience of which will be discussed in Part 3.
Advanced genetics has become a major tool for paleoanthropologists in the last few decades. The field also uses simpler methods to obtain data, such as observing how people live day to day, including what, when and how they eat. Hunter-gatherers existing in that last few hundred years as well as a handful of ones still in existence today have had their dietary habits recorded. Plainly speaking, looking at what modern-hunter gatherers ate allows us to build a picture of foods we evolved to eat. This will not reveal the whole story or even most of it, but it will give us some idea of which properties of certain foods our biology is adapted to exploit.
I will argue that observations made so far support the idea that low-carbohydrate diets were the most common kind and that ketogenic diets were less common but by no means unusual. Higher, moderate-carbohydrate diets were also quite common. On the other end of the spectrum, the late professor Staffan Lindeberg observed that there were healthy hunter-gatherers from the island of Kitava who consumed 65%-70% of their calories mainly from starchy tubers and fruit. With this broad overview in mind let’s put numbers to this so that we can refine our dietary picture.
In 2011, Ströhle and Hahn (4) looked at the ratio of plants-to-animals in the diets of 229 hunter-gatherer and horticulturalist groups. They combined these observations with measurements of food composition from Australian Aboriginals diets. This enabled them to infer the carbohydrate content of a particular group’s diet. Tables 3 and 4 below contain their data see please dive into it. I summarize their major findings as follows:
- A lower-carbohydrate, higher-fat diet is the most common sort of diet. Specifically, 16%-22% dietary carbohydrate is the most frequent (median and mode) macronutrient apportioning for 32.8% of groups. The 2nd most frequent is 29%-34% dietary carbohydrate for 27.9% of groups.
- Where hunter-gatherer groups are located geographically strongly correlates with how much dietary carbohydrate they consume. Specifically, latitude intervals strongly correlate, although not linearly, with the percentage of carbohydrate in the diet. 11° to 40° North or South of the equator, dietary carbohydrate accounts for 30%-35% of calories on average whilst a sharp decrease to 20%-9% occurs from 41° to 60° above or below the equator.
- Nearly 9 out of 10 of the diets of hunter-gatherer groups had less than a third of calories coming from carbohydrates. Specifically, “most hunter-gatherer diets (approximately 85%) were characterized by a relatively low carbohydrate intake (<35% of the total energy), which reflected the high reliance on animal-based foods of most hunter-gatherer societies”
- In the last quarter of our evolutionary timeline we started living higher up on the globe and thus had to switch to an even lower-carbohydrate diet. I am choosing the timeline of anatomically modern humans arising approximately 200,000 years ago. So specifically, “the switch to a low-carb diet (<25% of the total energy) seems to have taken place late in human evolution (ie, between 46 000 and 7 000 years ago) when our ancestors settled in higher latitude environments”.
- The authors made the assumption that “gathered food included only plant foods” which significantly overestimates the amount of carbohydrates in the diet given that non-plant foods includes “the collection of small fauna (eg, invertebrates, insects, and eggs)”.
What bones contain can tells us something about what we ate
I will tell you what isotopes are in a minute. Having just considered dietary inferences derived from a combination of observations and empirical food data, now let’s consider dietary inferences derived from empirical data obtained from looking at the isotope signature of human and other animal bones. In 2009 Richards and Trinkaus compared the isotopic signatures of bones from Neanderthals, various early modern Europeans including Oase 1, modern Europeans, canids (wolves), cervids (deer) and hyenas. Before we dig into the findings, it is important to understand their reliable method for looking back into our dietary past using isotope analysis. An isotope is the variant form of an atom (or element). Take Nitrogen for example. The most common form is stable and has 7 protons + 7 neutrons, earning the name 14N. One variant of Nitrogen (or isotope) has 7 protons + 8 neutrons, earning the name 15N. The unequal number of protons and neutrons makes the atom more or less heavy compared to one with equal numbers. The atom can be ‘weighed’ by a device called a mass spectrometer. By looking at the ratio of 12Carbon (12C) to 13Carbon (13C) in a bone sample for example, I can be reasonably confident about what sort of food that land animal ate given that some animals eat C3 plants (like tubers, fruits, nuts) and others C4 plants (like maize, millet, sorghum). C3 and C4 plants refers to how these plants fix carbon during photosynthesis by joining 3 or 4 carbon atoms together, respectively.
Now that the physics underpinning isotope signatures is out of the way, lets look at what we’re particularly interested in, the diets of 2 kinds ancestral humans. The main reason is because their timelines overlap, making the comparison more relevant. The first human, a 37,000 – 42,000 year old modern male called Oase 1 found in Peştera cu Oase (Romania) in 2002. Peştera cu Oase means “The Cave with Bones”! The Neanderthal skeletons used in this study were dated to less than 50,000 years ago on average. Although not immediately relevant, I can’t help but mention that in 2015, Fu et al. (5) discovered that an ancestor of Oase 1 had sex with a Neanderthal 4 to 6 generations before his own. So what does the data indicate overall? I’ve summarized it below.
- Neanderthals ate protein (meat, basically) mostly from land animals and probably very little to none from the sea. How much? “Each Neanderthal had δ15N values that were 3 to 5% higher than contemporary herbivores and similar to carnivores (or in some cases slightly higher)”. They are thus called “top-level” carnivores.
- The Oase 1 human ate lots of protein from somewhere else, most likely freshwater fish.
- Other early modern humans also ate lots of protein from the sea. How much? “[…]they consumed significant (i.e., 20–30% of dietary protein) amounts of marine foods, probably high trophic level (carnivorous) fish or marine mammals.”
- Depending on the early modern human, some ate lots of land and marine animals, maybe even including a higher proportion of those meats in their diet than the “top-level” Neanderthal carnivores did. How much? Their δ13C values “indicate that their protein came from terrestrial C3 (or freshwater) foods, yet many of them have high δ15N values, at or above the highest Neanderthal values.”
- Although the study has robust conclusions confirmed elsewhere it has limitations. First and most importantly, “the method only measures protein intake, many low-protein foods that may have been important to the diet (i.e., high caloric foods like honey, underground storage organs, and essential mineral and vitamin rich plant foods) are simply invisible to this method”. Secondly, it is hard to confirm the accuracy of certain comparisons because “comparative faunal isotopes” were sometimes lacking. Lastly, “carbon and nitrogen isotope values vary between different geographical regions, especially related to temperature and aridity”.
- Humans ate a “wide range of diets”. This does however come with 3 notes of caution. First, this does not encompass vegetarianism or veganism. I cannot emphasize this enough. Secondly, ‘variety’ here refers mainly to the origin of animal sources. All of the different human ancestors analyzed in this study were “top-level” carnivores. Third, it is correct but incomplete to state that humans are carnivores. Here, the term ‘carnivore’ refers to where humans fit hierarchically in the food chain. It is not a claim about a supposed inability to derive nutrition from plants. Humans are in fact obligate carnivores, in the sense that we did not and still cannot survive without some animal foods. Furthermore, most of the paleoanthropological evidence points to us being omnivores. My guess is that we are preferential omnivores to varying degrees according to what the climate and fauna allows. Depending on the developing (‘primitive’) group, their particular culture may have influenced their position on the spectrum of carnivory to omnivory.
Finally, lets visualize the data used for the above conclusions
In Figure 1, δ13C values below “-19” indicates a diet of primarily or only land animals. This is the case for Neanderthals (empty squares) but not for some early modern European humans (filled circles). In Figure 2, you can see that the Oase 1 human has higher δ15N values than some carnivorous wolves! High δ15N values occur in this instance because when a predator eats its prey, there is a 3%-5% δ15N enrichment. Consequently, carnivores such as wolves for example, will have higher δ15N values than their prey (such as ibex).
The takeaway here is that observations made of modern hunter-gatherers and clues from ancestral human bones support the claim that humans commonly evolved on low-carbohydrate and sometimes ketogenic diets. Seasonal changes in food availability likely swung the pendulum between higher-carbohydrate all the way down to ketogenic. In this sense, it is not correct to say that ketogenic diets are extreme diets in terms of their occurrence amongst humans.
In part 3, we will discuss how a distinguishing feature of humans – our big brains – is all the rage in hominization theories and how a high-fat diet may have been an important driver in it. However, the question is far from settled.