Thursday, May 26, 2011

Who Said Paleo Diet Was High In Fat? Part 2.0

In 2000, a team including Loren Cordain, Janette Brand Miller, S Boyd Eaton, Neil Mann, Susanne HA Holt, and John D Speth published “Plant-animal subsistence ratios and macronutrient energy estimations in worldwide hunter-gatherer diets” (hereafter referred to as “the Plant-Animal Ratios paper”) in the American Journal of Clinical Nutrition [1].  Based on their analysis of the diets of 229 tribes that they identified as contemporary hunter-gatherers, they concluded:

“Whenever and wherever it was ecologically possible, hunter-gatherers would have consumed high amounts (45–65% of total energy) of animal food. Most (73%) hunter-gatherer societies worldwide derived >50% (≥56–65%) of their subsistence from animal foods, whereas only 13.5% of these societies derived more than half (≥56–65%) of their subsistence from gathered plant foods. In turn, this high reliance on animal-based foods coupled with the relatively low carbohydrate content of wild plant foods produces universally characteristic macronutrient consumption ratios in which protein intakes are greater at the expense of carbohydrate.”[Italics added.]

Specifically, Cordain et al decided that:

“The most plausible (values not exceeding the mean MRUS) percentages of total energy [in hunter-gatherer diets] would be 19–35% for dietary protein, 22–40% for carbohydrate, and 28–58% for fat.”

Although this suggests that hunter-gatherer diets ranged from low- to high-fat (28-58% of energy from fat), this paper has served as support for the idea that hunter-gatherer diets typically had high fat contents, because in it Cordain et al appear to provide evidence for three propositions:

1.  Hunter-gatherers would have eaten more animal than plant food “wherever and whenever possible;”
2.  Wild plant foods have a “relatively low carbohydrate content;” and
3.   The paucity of plant foods available to hunter-gatherers made it necessary for them to focus their hunting on procuring large game supplying large amounts of fat. 

So, does this paper supply good evidence for these propositions? 

Perhaps unbeknownst to many people familiar with this paper, Katherine Milton, professor of physical anthropology at U.C. Berkeley, and author of at least 100 peer-reviewed papers on the dietary ecology of primates, human ancestors, and humans, including “Diet and Primate Evolution [pdf]” in Scientific American  and “A Hypothesis to Explain the Role of Meat-eating in Human Evolution [pdf]” in Evolutionary Anthropology, wrote a critical editorial commentary on the Cordain et al paper (Hunter-Gatherer Diets--A Different Perspective) that appeared in the same issue of AJCN, and a follow-up letter to the editor.  What did she have to say?


Cordain et al reported the procedure used to produce their numbers:

First, they analyzed the data from Murdock’s Ethnographic Atlas (heareafter referred to as the EA) to determine the probable range of plant:animal ratios in hunter-gatherer diets.  From this analysis they concluded that most hunter-gatherers derived more than half (56-65%) of their subsistence from animal foods.

Second, using a database of 829 wild plant foods, they calculated that the (mythical) “average” wild plant food macronutrient provided 62% of energy from carbohydrate, 24% from fat, and 14% from protein, and an energy density of 6.99 kJ/g (1.7 kcal/g).  

Third, unlike Eaton et al, Cordain et al assumed that hunters would consume all edible portions of a wild game carcass, including marrow, adipose, and visceral fats. 

Fourth, they assumed a constant of ~35 percent of the diet of all hunter-gatherers consisted of land animal foods, with any additional animal foods coming from fresh- or salt-water fish. 

Fifth, they state:

“Because of the similarity (3.5% difference) in the mean energy density of wild plant (6.99 kJ/g) and animal foods (7.24 kJ/g) in our database, we assumed that the P-A subsistence ratio based on weight in the Ethnographic Atlas would be virtually identical to the P-A subsistence ratios based on energy.”

Sixth, they assumed that an 80 kg hunter would require about 3000 kcal (12552 kJ) from food each day.

Seventh, they performed calculations using these numbers to generate ranges of macronutrients in a hunter-gatherer diet.

Eighth, if any calculation produced a protein intake greater than 35 percent of energy, which would probably deliver more nitrogen than the human liver can covert to urea in a day and therefore lead to harmful levels of blood ammonia and amino acids,  they assumed that ‘typical’ hunter-gatherers pursued higher intakes of animal fats. 

They based this assertion on the ethnographic reports in the EA and optimal foraging theory.   In short, from the data in the EA it appears that hunters preferred to increase animal fat as opposed to plant food consumption, even to the extent of selectively eating animal fats (and discarding the rest of the carcass).  Cordain et al explain this as a manifestation of optimal foraging theory, i.e. hunting a large fat animal supposedly has a greater return on investment than plant foods, meaning that for hunting fat animals, the ratio of energy obtained to energy spent in the hunt appears greater than the ratio of energy obtained to energy spent in collecting wild plants.

Thus, the conclusions made by Cordain et al rest entirely on the reliability of the basic data in the EA as a basis for making quantitative calculations about macronutrient ratios of hunter-gatherers.

Reliability of the EA Data

Cordain et al describe the nature of the EA data in this passage:

“Although Murdock did not specify whether the subsistence-dependence categories were based on the energy content or weight of the food for each subsistence economy (gathered plant foods, hunted animal foods, and fished animal foods), examination of the >400 original references indicates that in many cases, estimates were made by weight. Ethnographic data are qualitative in nature and as such lack the precision of quantitative data; consequently, Murdock's subsistence-dependence categories, in almost all cases, represent subjective approximations by Murdock of the ethnographer's or anthropologist's original observation.” [Italics added]

In this last sentence, Cordain et al appear to admit that the Ethnographic Atlas reports Murdock’s “subjective approximations” of the subsistence ratios of various cultures, based on the original, first hand observations of ethnographers or anthropologists, which are “qualitative” [i.e. plant versus animal], not “quantitative,” and “lack precision.” 

So the question arises, how can anyone have confidence in quantitative descriptions of the macronutrient ratios of hunter-gatherer diets when the basic “data” supporting those consists of second-hand “subjective approximations” of the raw weights of plant, animal, and fish consumed by the various tribes?

In discussing the limitations of their calculations, Cordain et al return to the issue:

“Perhaps the most important variable influencing the estimation of the dietary macronutrient ratio in hunter-gatherer populations, when indirect procedures are used, is the validity of ethnographic data. Other ethnographers who compiled hunter-gatherer data from the Ethnographic Atlas noted that the scores Murdock assigned to the 5 basic subsistence economies are not precise, but rather are approximations (11, 36, 37) generally based on raw weights of the dietary items (36). Although estimations of energy by weight of wild plant and animal foods may sometimes yield results similar to actual values, there is considerable room for error. The present analysis indicates that if the mean plant-food energy density for 829 wild plant foods (6.99 kJ/g) is contrasted with the energy density (7.24 kJ/g) of an average white-tailed deer with 10% body fat, there would only be a 4% difference between actual energy values and those estimated by weight. However, if the mean energy density of wild fruit (3.97 kJ/g) or wild tubers (4.06 kJ/g) were contrasted with that of a white-tailed deer with 17.7% body fat (10.17 kJ/g), there would be a 60–61% difference between actual energy values and those estimated by weight. Obviously, not all ethnographic estimations of energy intake in hunter-gatherer populations based on food weight would necessarily be this extreme. This example does indicate the imprecise nature of qualitative ethnographic data; however, it does not rule out its important use as a data source to test hypothetic models. [Italics added.]”

Cordain et al admit that the EA data is “qualitative” and “imprecise,”  and admit "considerable room for error" involved in translating this data into energy and macronutrient intakes, yet all of their more precise quantitative conclusions about macronutrient ratios rest on this somewhat shifty foundation. 

Milton had a few things to say about this use of the EA.  In her critical editorial comments  on Cordain et al, Milton lists some of the reasons she doubts the accuracy of conclusions drawn from unselective use of the EA dietary data on "hunters:"

1 The EA is “compiled largely from 20th century sources and written by ethnographers or others with disparate backgrounds, rarely interested in diet per se or trained in dietary collection techniques.”  These people would not compile accurate objective data about diet but would give their  subjective impressions.
2  “By the 20th century, most hunter-gatherers had vanished; many of those who remained had been displaced to marginal environments.”   Marginal environments may produce less edible plant matter than richer environments favored by agriculturalists; if given a environment richer in plant resources, they might choose something else.  For example, Inuit are forced to live on an animal-based diet because their environment does not supply calorically significant plant foods.  This doesn’t tell us anything about what they would do if they inhabited the lusher environments now under the control of agricultural people.   Similarly the contemporary !Kung live in a desert; what they eat now doesn’t tell us what their ancestors ate when the Khalahari was a lush land with many large permanent lakes.
3  “Some societies coded as hunter-gatherers in the Atlas probably were not exclusively hunter-gatherers or were displaced agricultural peoples.”  More about this below.
4  “Because most of the ethnographers were male, they often did not associate with women, who typically collect and process plant resources.” 
5  “Finally, all the hunter-gatherers … included in the Atlas were modern-day humans with a rich variety of social and economic patterns and… not ‘survivors from the primitive condition of all mankind.’ (6). Their wide range of dietary behaviors does not fall into one standard macronutrient pattern that contemporary humans could emulate for better health.”

All of these factors introduce considerable room for error in the EA representations of hunter-gatherer diets, several producing a bias toward an impression that hunter-gatherers favored animal over plant foods. 

In her letter to the AJCN that also accompanied the Cordain et al paper, Milton made further remarks:

“In his 1968 analysis of hunter-gatherer diets, [Richard] Lee (8) reclassified some Atlas data and also excluded mounted hunters with guns and ‘casual’ agriculturalists from his database. In Lee's opinion, only 24 societies from all of Africa, Asia, Australia, and South America could be classified as hunter-gatherers, whereas North America alone contained >80% (135) of the 165 ‘hunting’ societies listed in the Atlas.”
“In contrast, in their analysis, Cordain et al (1) identified 229 hunter-gatherer societies in the Atlas; they also combined 2 of Lee's discrete categories (hunting and fishing) to estimate the total contribution of animal foods to energy subsistence. Given the uneven quality of most dietary data in the Atlas, the overrepresentation of hunter-gatherer societies from more temperate locales and the differences in classification and data analysis between these authors, different conclusions seem inevitable and all conclusions appear to merit closer study.”

Richard Lee provided his opinion in his essay “What hunters do for a living, or, how to make out on scarce resources” published in the book he edited, Man the Hunter (1968).  According to the University of Toronto Archives and Records:  “Prof Lee has published over 100 articles and chapters in books. He has authored several books including Man the Hunter (1968), Kalahari Hunter Gathers (1976), Politics and History in Band Societies (1982) and The Dobe Ju/’hoansi (2003). Most recognized is his 1979 , The !Kung San: Men and Women and Work in a Foraging Society, listed in American Scientist list of the 100 most important works in science of the 20th century.”

Now, consider Milton’s points: 

1.   Lee came to the conclusion that only 24 societies in the EA could be legitimately classified as hunter-gatherers operating in truly primitive fashion, i.e. without firearms, horses, or “casual agriculture,” but Cordain et al classified as hunter-gatherer any tribe that, according to the “subjective approximations” of Murdock, did not practice animal husbandry or agriculture at the time of observation by ethnographers or anthropologists, even if that tribe used modern firearms, or went hunting on horses instead of on foot.  The use of firearms and horses certainly increases the productivity of hunting compared to hunting on foot with primitive bow and arrow, perhaps leading some tribes to hunt--and succeed--more than they would otherwise. 
2.  None of the tribes that Lee identified as hunter-gatherers inhabited North America, whereas 80% of the tribes classified as “hunters” in the Atlas inhabited mostly 20th century North America.  In the 19th and 20th centuries, Euroamericans forced Natives of North America off of productive, agriculturally valuable lands that could produce large amounts of plant biomass, onto marginal lands like highlands and deserts where they would have little opportunity to consume plants relative to animals.

If Milton is correct, this apparently means that Cordain et al considered North American mounted hunters using firearms more “representative” of hunter-gatherers than the pedestrian African !Kung and Hazda, which in this letter to the AJCN they describe as “extreme” and “unrepresentative” of hunter-gatherers because they have diets higher in plant food than those North American tribes. 

In reference to this latter idea, Milton responded:

“The !Kung and Hazda, dismissed by Cordain et al as ‘unrepresentative,’ differ from many hunter-gatherers listed in the Atlas precisely because they have been relatively well studied dietarily—in both cases, plant foods contributed the bulk of daily energy intake.”

So, in essence, Milton is asking, which serves as the better basis for estimates of evolutionary macronutrient intakes: Quantitative data derived from first-hand accounts of African tribes not using horses or firearms, or qualitative data derived primarily from second-hand “subjective approximations” of dietary intakes of tribes all over the world, but predominantly from contemporary North America?   

In "Stone Agers in the Fast Lane"  and The Paleolithic Prescription, Eaton et al used the plant:animal subsistence ratio estimated by Lee, based on the 24 tribes he identified as primitive hunter-gatherers, namely 65:35 plant:animal, as the basis for their macronutrient estimates.  

Enough for now.  Remember that Cordain et al claimed that wild plant foods typically have a "relatively low carbohydrate content," apparently making them unsuitable as staple foods for hunter-gatherers, making it necessary for them to hunt high fat animals to avoid excessive consumption of protein.  Both Milton and I have some observations on this claim, which I will discuss in a future post, part 2.1.


Tuck said...

Very interesting post, Don.

I think the key point is that our ancestors would have eaten a wide range of food, and been healthy. This would have varied from tribe to tribe, and around the year.

I think arguing that a high-fat diet is bad is as wrong as arguing that a high-carb diet is inherently bad. With adequate nutrition and avoiding food toxins, our bodies ought to be able to operate on either.

The problem with Cordain's data is that it's based on speculation, and the problem with the African HG tribes is that they live on marginal land eating marginal food.

Neither gives us an accurate picture of what a human would have eaten in Africa 50,000 years ago.

An important point to remember.

Stephan Guyenet said...

Hi Don,

While I agree with the main thrust of your post, the problem with Lee's original analysis is it considered fishing, shellfishing and small game hunting to be "gathering", therefore it was lumped into the "plant" category. So I think Lee's analysis was also incorrect. That's something Cordain corrected in his analysis of Murdock's atlas data, although Cordain's analysis had its own problems as you pointed out.

Also, I've never been very impressed by Milton's reasoning. She comes from the perspective that primates like chimps and gorillas are a good model for human nutrition. I think a cursory look at the chimp and gorilla diets should put that idea to rest, as they eat virtually no starch, a massive amount of fiber, and get a large proportion of their energy from colonic fermentation. Milton recommends a low-fat diet, but a low-fat diet for humans looks nothing like a wild chimp/gorilla diet.

Don said...


In their letter to the editor, Cordain et al wrote:

"on the contrary: as intakes of plant food decrease with increasing latitude, intakes of fished food increase and of hunted animal food stay constant."

In the original paper also, Cordain et al claimed that hunted food remained constant at 35% at all latitudes, and that in upper latitudes, fished foods replaced plant foods as the latter formed less of subsistence. Here's the quote from the original paper:

"Because previous analyses of the Ethnographic Atlas (11, 36) as well as the present analysis indicate that hunted animal food makes up ≈35% of the subsistence base for worldwide hunter-gatherers regardless of their resident latitude or environment, we used this constant figure. The remainder of the animal food, when the percentages of animal food exceeded 35% of energy, was assumed to be either fresh or saltwater fish. Consequently, in the present model, with decreasing animal-food (hunted + fished) intakes, fish-food intakes decrease as plant-food intakes increase and hunted-animal-food intakes remain constant. "

This implies that adding the fish to the animal food category may have only substantially affected the P:A subsistence ratio for tribes in upper latitudes, therefore summing hunted and fished foods would have had little or no effect on Lee's calculations, since he used only 24 tribes at lower latitudes (Africa, Asia, Australia, and South America), where, presumably, fished foods played little role (e.g. among the !Kung in the Khalahari...where would they get fish regularly anyway?).

I find it hard to say that Cordain et al "corrected" some error in Lee's estimates since they used a completely different set of tribes (229 vs 24) and ecological conditions, as well as more lax standards for identifying a tribe as "hunter-gatherer."

Poisonguy said...

Don, what was your opinion of Milton and her 100 "peer-reviewed" articles a couple month's ago? Did the word "quack" cross your mind?

Asclepius said...
This comment has been removed by the author.
Asclepius said...

Some interesting points Don. It is good to see someone challenging paleo ideas head on!

But given the glacial and interglacial periods of the Paleolithic, would plants really have been a reliable source of food for much of the time annually, never mind over the longer term?

Also, given the appearance of composite tools (spears/tools and so forth made of stone, wood or bone), and clothing, doesn't this suggest a big dependence on animals for much more than food and so make it more likely energy was focused on hunting rather than gathering? Animals would have been high value with high utility. Moreso than a plant don't you think?

Stephan Guyenet said...

Hi Don,

I take your point about the contribution of fish and shellfish-- I think it's clear that cultures like the Inuit and Pacific Northwest groups relied on it to an extreme degree relative to other cultures. Although I'm not sure it's safe to assume Africans weren't getting a significant amount of food from fish.

I still wonder about the contribution of small game "hunting" that would have been considered plant food by Lee. Things like collecting amphibians and insects.